| cehteh | hi, i need a hint, i updated a server to beowulf (was long overdue :D), now it drops me into the initramfs shell with no md devices visible | 00:41 |
|---|---|---|
| cehteh | anything changed from ascii -> beowulf? | 00:42 |
| plasma41 | cehteh: What method did you use for the upgrade? | 00:44 |
| cehteh | apt dist-upgrade / aptitude and some manual fixing because not everything gone well | 00:46 |
| cehteh | server is at hetzner, i am now in a rescue system | 00:47 |
| plasma41 | cehteh: What filesystem is in use? | 00:48 |
| cehteh | ext4 | 00:48 |
| cehteh | root partititon is on md0 : active raid1 nvme0n1p2[0] sdb2[2](W) sda2[1](W) | 00:48 |
| plasma41 | cehteh: Is this using a hardware raid controller or software raid? | 00:51 |
| cehteh | software raid | 00:51 |
| cehteh | self build kernel i am just checking if that missing anything (but i used a known working kernel config) | 00:52 |
| plasma41 | hmm.. I'm not very familiar with either nvme or software raid | 00:55 |
| cehteh | famopus words: it worked before :D | 00:55 |
| cehteh | ok its not the kernel, devuan kernel doesnt boot either | 01:07 |
| cehteh | hi doc :D | 01:09 |
| mason | cehteh: chroot in, rebuild initramfs? | 01:09 |
| cehteh | mason: i did already | 01:09 |
| mason | cehteh: Any error message as it ditches you in the initramfs shell? | 01:09 |
| cehteh | there are sme efi fuckups, nothing to worry about, otherwise nothing i see | 01:10 |
| cehteh | i suspect it builds the initrd somehow wong, but no idea what happens | 01:12 |
| cehteh | https://pbs.twimg.com/media/E5z2El5WUAYRgI1?format=jpg&name=4096x4096 | 01:13 |
| mason | cehteh: From rescue, delete your initramfs or move it aside and generate it fresh. | 01:13 |
| mason | I recently saw one consistently update wrong but do the right thing on a fresh build, even though I'd said 'update-initramfs -c' which ought to have started from scratch. | 01:14 |
| cehteh | the missing dm-raid45 comes from me installed the dmraid package (but i useing mdraid) | 01:14 |
| cehteh | lemme try | 01:14 |
| mason | Failing that, lsinitrd from rescue and see if anything obvious is missing. | 01:15 |
| cehteh | well no i am rebootin into the rescue system again and force-rebuild the initrd, moment | 01:16 |
| cehteh | ok in the new generated initrd everything is there, rebooting | 01:22 |
| cehteh | nope, md stuff doesnt come up | 01:25 |
| cehteh | is there a mdadm spell to rescan and discover md devices? | 01:26 |
| cehteh | from initrd lsmod shows all raid modules are loaded, it just somehow doesnt discover them | 01:28 |
| rrq | is eudev installed? | 01:30 |
| cehteh | yes | 01:30 |
| cehteh | is that part of initrd? | 01:30 |
| cehteh | and damn remote screengrabber/console i cant enter a - | 01:31 |
| rrq | yes .. it all sounds like a (e)udev issue | 01:31 |
| * cehteh wants to run a mdadm --auto-detect | 01:32 | |
| cehteh | but without able to type a - grr | 01:32 |
| rrq | are the /dev/md nodes there? | 01:32 |
| cehteh | nope | 01:32 |
| cehteh | i think that causes the problem | 01:32 |
| rrq | you have /dev/nvme* nodes? | 01:33 |
| cehteh | hey on a sitenote: can one install and start a sshd from initrd? | 01:33 |
| cehteh | yes all drives incl nvm are there | 01:34 |
| rrq | you should be able to mount the /dev/ndme0p2 .. or whatever was the root, and then chroot into it | 01:34 |
| cehteh | i can do that from the rescue system yes | 01:35 |
| rrq | grom the initramfs prompt as well | 01:35 |
| cehteh | but the remote console (screengrabber, web browser based) sux, i cant type all characters | 01:35 |
| rrq | from | 01:35 |
| cehteh | ah ... and the root is on raid | 01:36 |
| * rrq oops .. need to go | 01:37 | |
| cehteh | i could break the raid1 and somehow forcemount a component of that, but that wont solve anything | 01:37 |
| mason | nvme0n1p# | 01:39 |
| mason | namespaces | 01:39 |
| cehteh | ? | 01:40 |
| mason | cehteh: I just caught up on scrollback. So, you need to get the right mdadm.conf into the initramfs. That might not be the only problem, but it's necessary. | 01:40 |
| cehteh | note: it ran before, i didnt changed the disc/raid config and i can start the raid and mount everything from the rescue system | 01:41 |
| mason | cehteh: So, like, mdadm --detail --scan >> /etc/mdadm/mdadm.conf | 01:41 |
| mason | in your rescue chroot | 01:41 |
| mason | the edit that file and make sure there's no default/incorrect cruft | 01:41 |
| cehteh | the madm.conf is correct | 01:42 |
| mason | the update-initramfs -c -k all or however you like to do it, and it ought to know what it's looking for | 01:42 |
| cehteh | uuid based | 01:42 |
| mason | cehteh: Maybe try path-based in case it's having trouble finding the UUID? Random guess. | 01:42 |
| mason | That seems unliekly though. | 01:43 |
| cehteh | it should at least have the device nodes /dev/md* | 01:43 |
| cehteh | but there are none, no uuid/no name nothing | 01:44 |
| mason | cehteh: Maybe try explicitly adding them to /etc/initramfs-tools/modules ? Shouldn't be necessary, but it might be easier figuring it out from your booted system. | 01:45 |
| cehteh | the modules are there | 01:45 |
| cehteh | lsinitramfs showing them | 01:46 |
| cehteh | the scripts too, i just didnt figure out yet what the scripts are doing | 01:46 |
| cehteh | lsmod from initrd also shown that the modules are loaded | 01:47 |
| cehteh | it just doesnt scan/find any md devices | 01:47 |
| * cehteh never had that problem before | 01:47 | |
| mason | cehteh: Did you say you can construct them by hand? | 01:48 |
| cehteh | from the rescure system they are there | 01:49 |
| cehteh | autodetected | 01:49 |
| mason | If you can do it from the initramfs environment, you can mount up your root on... is it /sysroot? Get everything loaded by hand, and boot from there. | 01:49 |
| cehteh | from the initrd i cant because i cant type into that crapp screengrabber console | 01:49 |
| mason | Not from the rescue system, from the actual busybox. | 01:49 |
| mason | heh | 01:49 |
| cehteh | try to do shell stuff without the - character | 01:49 |
| mason | It's worth a try. mdadm --start --scan, lvm as needed, get the root into place and let the system pivot after you exit | 01:50 |
| cehteh | cant type - | 01:50 |
| cehteh | yes wtf | 01:50 |
| mason | That's weird. | 01:50 |
| cehteh | well its some screengrabber thingy thats remote over webbrowser support hooks that up | 01:50 |
| mason | Oh, oh, I see. | 01:50 |
| mason | I forgot that detail. | 01:51 |
| cehteh | plugging vga and keyboard in doing magic sauce | 01:51 |
| cehteh | the wrong way | 01:51 |
| cehteh | totally fucked up keyboard layout | 01:51 |
| cehteh | i tried all keys with shift and with AltGr .... no - | 01:52 |
| mason | If it's just the layout, can you run loadkeys and get a saner one? | 01:52 |
| cehteh | how to install keymaps into the initrd? | 01:52 |
| mason | looking | 01:53 |
| cehteh | loadkeys is there found that, but couldnt load 'de' :) | 01:53 |
| cehteh | ideally i would like a initrd that has a failsafe that fires up network and dropbear | 01:54 |
| mason | cehteh: Well. If there's a loadkeys binary, maybe you can construct an initramfs that contains enough to get you a hyphen anyway. | 01:54 |
| cehteh | getting rid of that crap screengrabber | 01:54 |
| mason | cehteh: You can construct one of those and drop it in, but dropbear ... been a while since I did that, and I stopped doing that as there was some random instability after I booted. | 01:55 |
| cehteh | i dont mind a bloated initrd ... i jsut dont know how to add such things there, i wish initrds would be minimal *real* systems | 01:55 |
| cehteh | yes i did something like that too, thre are some network fuckups in the kernel | 01:55 |
| cehteh | other story (you cant really deconfigure a network interface .. i mean *really*) | 01:56 |
| cehteh | putting it down isnt enough | 01:56 |
| cehteh | anyway thats another thing | 01:56 |
| mason | cehteh: So, I see a possible strategy for you. dumpkeys > foo, edit that, shim it into your initramfs. I've not built one by hand since before the started front-loading microcode, but then you'd have a keymap you could use with loadkeys and get a usable console. | 01:57 |
| mason | It's a lot of messing around for a console, but it might be useful. | 01:58 |
| cehteh | well the console can time out any time now (asked support for 2 hours, didnt expected so mcuh fuckup) | 01:58 |
| mason | Yeah, this is weird. You'll have to be creative with it. | 01:58 |
| mason | Hard to figure out at one remove and the kids are calling me to play a game, but if I think of anything I'll come back and mention it. | 01:59 |
| Harzilein | hi | 01:59 |
| cehteh | hi Harzi :) | 01:59 |
| * Harzilein peruses the logs | 01:59 | |
| cehteh | Harzilein: initrd doesnt pull up the md raid | 01:59 |
| cehteh | no idea why, all modules and stuff are there | 02:00 |
| cehteh | is there any special grub module needed for raid1? | 02:16 |
| gnarface | i don't think so but LVM and FDE can both cause various complications when used together with it | 02:18 |
| gnarface | also, in case you didn't know, just using it for / can cause complications | 02:19 |
| gnarface | it actually isn't recommended to boot from a raid partition | 02:19 |
| gnarface | you should put everything else on raid if you can afford it but the boot partition should have as few extraneous hardware requirements as possible | 02:20 |
| adhoc | whats FDE ? | 02:20 |
| gnarface | full disk encryption | 02:20 |
| adhoc | ok | 02:20 |
| adhoc | makes sense | 02:20 |
| adhoc | putting /boot on a file system more complex that EXT4 has always been an issue | 02:21 |
| gnarface | on low-powered ARM devices, even using ext4 really bogs down boot times | 02:21 |
| adhoc | just recently went through the whole dance of getting ubuntu server 20.04 LTS to boot off a pair of M2's with mdadm | 02:22 |
| adhoc | thankfully it allows you to configure that at install time | 02:22 |
| adhoc | is a pian to do that afterwards | 02:22 |
| adhoc | gnarface: ok | 02:22 |
| adhoc | gnarface: why is ext4 an issue with booting on Arm ? | 02:23 |
| gnarface | not sure but i think it's more about the performance constraints of flash and low-power flash drive controllers | 02:23 |
| gnarface | i observed that the problem doesn't even occur if booting from a ext2 partition using the ext4 driver in reverse-compatibility mode | 02:24 |
| adhoc | booting from flash, you have more appropriate filesystems | 02:25 |
| gnarface | (though it is more popular to use fat32 for /boot on such devices i think it is not a superior choice except if you think you need to repair your /boot partition contents from a stock windows install | 02:25 |
| gnarface | ) | 02:25 |
| adhoc | like the jffs, etc ? | 02:25 |
| gnarface | f2fs is the new one everyone is excited about | 02:25 |
| gnarface | but i don't know if anyone is using it for /boot yet either | 02:25 |
| adhoc | ok, how time change =) | 02:25 |
| gnarface | also, samsung contributed it so you know... probably loaded with backdoors | 02:25 |
| adhoc | is there a particular class of device that you are refering to ? | 02:25 |
| gnarface | well, i observed it across the board on the allwinner a64 stuff that pine64.org is selling | 02:26 |
| gnarface | phones and notebooks made from tablet guts | 02:26 |
| adhoc | ok | 02:26 |
| cehteh | i dont booting from raid1 since ages (close to 20 years perhaps) that always worked | 02:27 |
| gnarface | adhoc: but i think it is more about power margins than cpu performance | 02:27 |
| gnarface | adhoc: don't really know for sure | 02:28 |
| gnarface | cehteh: it's not that it doesn't work it's just that we don't advise it because it makes both installation and disaster recovery complicated, which is a bigger threat to the type of people who are asking for advice on this type of thing in the first place | 02:29 |
| cehteh | huh? | 02:29 |
| gnarface | cehteh: it's a pitfall for noobs | 02:30 |
| cehteh | actually grub has excellent support for that it can boot even when the initial boot disc is defect when properly set up | 02:30 |
| gnarface | cehteh: and it doesn't buy you much because the /boot partition shouldn't be getting read or written much at all under normal useage | 02:30 |
| cehteh | i am programmer not admin, but wont call myself noob | 02:30 |
| cehteh | it buys you that it boots even when the disc is defect | 02:31 |
| gnarface | if it's not raid0 | 02:32 |
| cehteh | md0 : active raid1 nvme0n1p2[0] sda2[2](W) sdb2[1](W) | 02:32 |
| gnarface | if you know what you're doing then the warning doesn't apply to you | 02:32 |
| cehteh | well i *mostly* thought i know what i am doing .. and it worked on many machines until this botched update | 02:33 |
| cehteh | and i have zero clue whats wrong now | 02:33 |
| cehteh | initramfs is already linux, thus grubt could boot from raid1 .. | 02:37 |
| cehteh | just somehow linux doesnt pull the raid up .. but i may check the mdraid conf again | 02:39 |
| cehteh | .. its correct | 02:41 |
| gnarface | would you need to do something to push the config change into the initramfs? | 02:43 |
| gnarface | or did you already? | 02:43 |
| cehteh | nothing changed | 02:44 |
| gnarface | wait, it just stopped working between reboots? | 02:44 |
| cehteh | its the same config since 5b years or so | 02:44 |
| gnarface | i thought you updated or something | 02:44 |
| cehteh | yes updated from ascii to beowulf | 02:44 |
| gnarface | oh, so the update broke it | 02:44 |
| cehteh | but the raid config stayed the same | 02:44 |
| cehteh | yes | 02:44 |
| gnarface | hmmm, disturbing | 02:44 |
| gnarface | doesn't have a nvidia card in it, does it? | 02:45 |
| cehteh | its a server! | 02:45 |
| gnarface | well, people do crazy things | 02:45 |
| gnarface | lol | 02:45 |
| cehteh | yes i dont mine bitcoins | 02:46 |
| rrq | cehteh: what does the md conf file say? does it mention the nvme partitions? | 02:46 |
| cehteh | its uuid based | 02:46 |
| cehteh | but moment .... | 02:46 |
| cehteh | 2 1000MB 32.0GB 31.0GB ext4 root legacy_boot | 02:46 |
| cehteh | .. thats the partition ... shouldnt that have the 'raid' attribute set for autodetection? or was that obsoleted? | 02:47 |
| cehteh | ^ parted output | 02:47 |
| cehteh | set that, retrying | 02:49 |
| rrq | isn't there an /etc/mdadm.conf declaring how that raid is assembled? | 02:49 |
| gnarface | it's not just like, slowly rebuilding in the background or something, is it? | 02:50 |
| cehteh | in old times the kernel could scan for 'raid' partitions and assembled them automatically | 02:50 |
| cehteh | the /dev/md nodes are not created and it falls into (initramfs) rescue | 02:51 |
| * rrq admits not using md myself | 02:51 | |
| cehteh | https://pbs.twimg.com/media/E5z2El5WUAYRgI1?format=jpg&name=4096x4096 | 02:51 |
| cehteh | and no the raid attribute didnt fix it | 02:51 |
| cehteh | i tought they obsoleted that .. anyway was worth a try | 02:52 |
| rrq | module dm-raid45 ? | 02:55 |
| cehteh | irrelevant | 02:55 |
| cehteh | after the problem started i installed dmraid stuff (despite i am useing mdraid) .. rebuilding the initrd pulled some things in, but the kernel is build without dmraid4/5 | 02:56 |
| rrq | you've been at https://askubuntu.com/questions/1219339/mdadm-error-opening-dev-md0-no-such-file-or-directory I suppose | 03:00 |
| cehteh | ok tried to boot a stone age kernel .. same problem it really lies somewhere in the initrd | 03:02 |
| rwp | Just started reading the scrollback... | 03:02 |
| rwp | cehteh, Do you have the option to boot from an arbitrary ISO such as the installer image? | 03:03 |
| cehteh | the rescue system boots from PXE | 03:03 |
| rwp | If so then you could use it's rescue mode to bootstrap the system and then use that to fix whatever is happening now. | 03:03 |
| rwp | What system boots from PXE? | 03:03 |
| cehteh | the rescue! | 03:03 |
| rwp | And it is a Devuan/Debian rescue image? Excellent! I did not expect that. I expected something else from the hosting company. | 03:04 |
| cehteh | its in a datacenter, normallly it boots locally but i can enable rescue then a rescuesystem is pushed via pxe | 03:04 |
| cehteh | there the md device are available | 03:04 |
| cehteh | if i would know whats wrong i could fix it | 03:04 |
| cehteh | but i have still absolutely no clue whats wrong | 03:04 |
| rwp | The installed system is more rigid in that it needs to know the mdadm.conf checksums to assemble the RAID. But the rescue system is more loose and ad-hoc about dynamically assembling what it finds there. | 03:05 |
| cehteh | did that change in devuan? | 03:05 |
| rwp | First, I don't know... But I read that you are using a custom kernel? Therefore I perhaps would try installing a stock kernel and seeing if it will boot successfully. | 03:06 |
| cehteh | the config is correct | 03:06 |
| cehteh | i tried stock kernel same problem | 03:06 |
| gnarface | what if you just run "update-initramfs -u" once just for good measure? | 03:06 |
| cehteh | did update-nitramfs -c -k all | 03:06 |
| gnarface | hmmm | 03:06 |
| rwp | Change? AFAIK the /etc/mdadm/mdadm.conf file is configured into the initramfs and must match. | 03:06 |
| cehteh | yes it matches | 03:06 |
| rwp | But in rescue mode one has the option to dynamically assemble all of the automatically found components. Which is usually exactly what is wanted. | 03:07 |
| cehteh | it also matches with mdadm --examine --scan | 03:07 |
| rwp | Also, GRUB has changed recently. I learned a lesson about how grub is configured due to a failed raid boot problem here. | 03:07 |
| cehteh | so the raid arrays are clean, fine, well configured | 03:07 |
| rwp | The grub-pc packaging (ab)uses debconf as a registry. It expects the boot drives to be listed in the debconf registry. (I don't know about grub efi packages but think they don't.) | 03:08 |
| cehteh | grub boots from raid (it loads the initrd and start is) | 03:08 |
| cehteh | with the initrd the mess begins | 03:08 |
| rwp | Learning that I went around to each of my grub-pc installed systems (ignored the efi ones) and "dpkg-reconfigure grub-pc" in order to fix them up. | 03:08 |
| cehteh | i guess i unpack the initrd and check whats exactly inside | 03:09 |
| cehteh | shall i try? dpkg-reconfigure grub-pc | 03:09 |
| rwp | I also know that the rescue image has all modules available. Or at least most. But installed system *might* have been configured more space conserving. | 03:09 |
| rwp | If you have grub-pc installed then that seems like a non-harmful option. | 03:10 |
| cehteh | i lsinitramfs'ed it, all md stuff is there, sure something may be missing but i have no clue what | 03:10 |
| rwp | None of us know exactly either. All we can do is try to brainstorm with you. | 03:11 |
| cehteh | ok installed grub on all discs (nvm and 2x spindle) | 03:12 |
| rwp | I will note that I have upgraded several systems from Debian Stretch up through into Devuan Beowulf all running RAID1 with mdadm software raid okay. | 03:12 |
| cehteh | yes i usually run that setup as well, sometimes debian sometimes devuan, all working | 03:12 |
| rwp | What that debconf registry does is that upon package upgrades later it will grub-install in the postinst to those registered locations. | 03:12 |
| cehteh | never had that problem before | 03:12 |
| rwp | So if those locations change then one needs to reconfigure the debconf registry for them. Otherwise grub won't be automatically updated on those devices. | 03:13 |
| cehteh | debugging initrd is always such a mess .. but this is really special cone is have no clue at all | 03:13 |
| rwp | And if grub upgrades the different phases and but does not upgrade the installed parts things get out of sync. Errors occur. | 03:14 |
| cehteh | ok problem stays after reconfigure | 03:14 |
| rwp | /me scrounged around in my bag and found this example error I have seen: error: symbol not found: 'grub_divmod64_full' | 03:15 |
| cehteh | reading through dmesg in initrd .... | 03:16 |
| rwp | If this were -offtopic I would mention that RHEL/CentOS includes the UUIDs to assemble on the kernel command line and that if not correct then odd behavior occurs... | 03:17 |
| cehteh | i use uuids too | 03:18 |
| cehteh | well dmesg shows that it probes the raid6 algos .. but nothing about trying to assemble the raid1 / boot device | 03:19 |
| gnarface | certain situations can trigger a UUID change though | 03:19 |
| gnarface | i assume those have all been verified to still be accurate? | 03:19 |
| cehteh | i am pretty confident that a reboot isnt one of these | 03:20 |
| gnarface | no, it would have had to be a reformat or resize i think | 03:20 |
| cehteh | nothing of that happend | 03:20 |
| adhoc | the UUID should be set at file system creation time ? | 03:20 |
| rwp | Since this seems like a module missing from initramfs I might also ask if /etc/initramfs-tools/initramfs.conf says something other than MODULES=most there? | 03:20 |
| cehteh | booting into rescue moment | 03:21 |
| rwp | For mdadm the UUID is set when the mdadm command creates the raid. | 03:21 |
| cehteh | and iirc it is most and i check lsinitramfs before all modules are there | 03:21 |
| cehteh | also in the botched boot (initramfs) shell lsmod shows all md modules loaded | 03:22 |
| rwp | Is this just mdadm raid or does it also include LVM on top of the raid? (All of mine are LVM on top of mdadm raid.) | 03:23 |
| cehteh | it just doesnt discover and start the arrays | 03:23 |
| rwp | I saw questions in the scrollback about /boot. Is that on the root partition or on a separate partition? | 03:23 |
| cehteh | the root partition (incl /boot) is just an raid1 nvm + 2 disks (with write-mostly flag) | 03:24 |
| rwp | I always put /boot on a separate raid /dev/md0 mount point. | 03:24 |
| cehteh | yes | 03:24 |
| cehteh | there is a 32gb root raid1 .. and a fat /data which is set up a bit more complcicated with dmcache .. but thats not my worries its not needed for booting | 03:25 |
| rwp | Hmm... So you are stuck in that crazy place where you can rescue and then chroot in and mount -a and start everything and it is running but it won't do it at boot time. Hmm... | 03:25 |
| cehteh | yes | 03:26 |
| cehteh | well rescue us some other system | 03:26 |
| cehteh | otehr kernel too | 03:26 |
| cehteh | but i also tried to boot the old kernel which definately worked .. but did not now | 03:26 |
| rwp | Well... The good news is that in that case you can get things running enough to sleep on the problem and attack it fresh in the morning. That's always good. Sleep is good. | 03:26 |
| cehteh | 03:26 | |
| cehteh | root@rescue ~ # cat /etc/debian_version | 03:26 |
| cehteh | 10.9 | 03:26 |
| cehteh | the rescues is a debian | 03:27 |
| cehteh | lol | 03:27 |
| cehteh | i dont want to put it live from the rescue system | 03:27 |
| rwp | I don't think there would be any problem using the Debian rescue versus Devuan rescue. | 03:27 |
| cehteh | also its not *that* important (some people may complain, like Harzilein ;) ) .. but its a private server no commercial stuff | 03:28 |
| cehteh | the thing now is that i cant sit that out thinking maybe i get an idea, i am absolutely clueless, i have to find some way to debug this | 03:28 |
| cehteh | tried the devuan distro kernel, same problem btw, its not a kernel problem | 03:29 |
| cehteh | its somewhere in the initrd | 03:29 |
| cehteh | so i just unpack the initrd and investigate whats going on there | 03:29 |
| rwp | It's a pretty unusual interesting problem. And if the names I see helping in the above can't think of a way out then it's a tough one. | 03:29 |
| rwp | I think you may need to learn more about building a custom initramfs to debug your problem than you expected to at this moment. | 03:30 |
| rwp | Since the initrd is being run (apparently) you have a starting point to try to figure out why it isn't starting the sw raid. | 03:31 |
| cehteh | yes this sux initrd problems always suck didnt happen to me for a long time | 03:31 |
| cehteh | i'd wish one could make the initrd do some terse logging with some simple commandline option, is there any? | 03:33 |
| cehteh | ideally to some file for later investigation | 03:33 |
| rwp | I do not know what I don't know about initrds but I imagine anything is possible there. It's running code. And that opens up a lot of possibilities. | 03:33 |
| rwp | If it were me I would grab a spare box here, install, then build up a custom initrd and debug my debug locally. Everything is easier locally. | 03:33 |
| rwp | Then after getting an understanding of things transfer that to the server. Use that to debug the current problem. | 03:33 |
| cehteh | yeah problem is that this is a remote server and should be up soonish again | 03:37 |
| cehteh | and i bet i couldnt replicate the problem | 03:38 |
| cehteh | i can diff the old initrd from backup before the fuckup with the current one | 03:38 |
| cehteh | still thats prolly a painful task because there are lots minimal details that may cause the problem | 03:38 |
| rwp | "Fast is slow. Slow is fast." | 03:38 |
| rrq | cehteh: apparently mdadm has its initrd script at /usr/share/initramfs-tools/hooks/mdadm .. | 03:44 |
| rrq | .. and it includes the ability to "auto generate" the conf | 03:44 |
| cehteh | yeah i know that | 03:44 |
| cehteh | its a bit old but i once looked into how initrd works | 03:45 |
| rrq | and does "/usr/share/mdadm/mkconf generate" create the same conf that you have? | 03:45 |
| cehteh | huh unpacking the initrd doesnt work, how is it compressed by default? | 03:51 |
| rrq | how about failing line commenting out line 53 in /usr/share/initramfs-tools/scripts/local-block/mdadm, the re-update initrd ...? | 03:51 |
| rrq | (the line with "mdadm -q --run /dev/md?*") | 03:53 |
| cehteh | i am still struggling to unpack the initrd .. haha | 03:53 |
| rrq | you'd edit on the host from where you pack the initrd | 03:53 |
| cehteh | yeah but i want to unpack it just to confirm its contents | 03:54 |
| rwp | gunzip -d or zcat the image into cpio and unpack it. zcat initrd | cpio -ivt | 03:54 |
| cehteh | yes so i thought | 03:55 |
| cehteh | # zcat /mnt/boot/initrd.img-5.10.47ct | cpio -vi | 03:55 |
| cehteh | gzip: /mnt/boot/initrd.img-5.10.47ct: not in gzip format | 03:55 |
| rwp | cpio uses -i to set input of the archive, -t is table of contents. cpio -idvm or something to create directories, verbose, modification times. But probably wrong as that was off the top of my head. | 03:55 |
| cehteh | well gzip already fails | 03:56 |
| rwp | zcat might look at suffix and expect a .gz. Try it with zcat < /mnt/boot/initrd.img-5.10.47ct | 03:56 |
| cehteh | nope | 03:56 |
| rrq | maybe use "file /mnt/boot/initrd.img-5.10.47ct" to see how it's compressed | 03:56 |
| cehteh | tried that, says cpio .. but thats wrong unpacking shows on ly a feaction of its contents | 03:57 |
| rwp | Works here. My Beowulf systems all have a gzip compressed initrd. | 03:57 |
| fsmithred | you have microcode installed? | 03:57 |
| cehteh | COMPRESS=gzip | 03:58 |
| rwp | I also note that "find . -print | cpio -o -H newc | gzip" is how initrd files would be created and compressed. It's that specific cpio format. | 03:58 |
| cehteh | yes | 03:58 |
| rwp | If you can't unpack the initrd and the initrd is failing then is that perhaps part of the problem? | 03:59 |
| cehteh | maybe | 03:59 |
| rrq | the initrd might be a concatenation of 2 parts microcode+filesystem | 03:59 |
| cehteh | *but why* | 03:59 |
| rwp | It's okay to concatenate gzip'd parts of initrd files together. | 03:59 |
| rwp | As long as all of them are cpio -H newc type and all gzip'd the same. | 04:00 |
| cehteh | # file /boot/initrd.img-5.10.47ct | 04:00 |
| cehteh | /boot/initrd.img-5.10.47ct: ASCII cpio archive (SVR4 with no CRC) | 04:00 |
| cehteh | i am puzzled | 04:00 |
| cehteh | lsinitramfs shows me a lot content | 04:01 |
| fsmithred | yeah, you have microcode installed | 04:01 |
| fsmithred | (cpio -i ; zcat | cpio -i) < "$initrd_image" | 04:01 |
| cehteh | ah | 04:01 |
| cehteh | yes microcode is in front i see | 04:01 |
| fsmithred | you can thank rrq for that line | 04:02 |
| cehteh | yes now i see | 04:02 |
| rrq | I think that mdadm scripts would be at /scripts/local-block/mdadm in the initrd | 04:09 |
| fsmithred | rrq, I didn't read all the scrollback, but is this possibly related to what jaromil was talking about with grub? | 04:10 |
| cehteh | ok ... i am a bit puzzled about UUID's | 04:11 |
| fsmithred | I avoid them for that reason | 04:11 |
| rrq | well looks to me like poor scripting in that mdadm script | 04:11 |
| cehteh | maybe i swith for labels | 04:12 |
| gnarface | there was some other situation they can change in that surprised me. it wasn't just resizing and reformatting, i think i ran into a situation where the UUID changes if the fsck job has to repair errors on the filesystem or something too? the memory is vague now .... | 04:12 |
| rrq | the failing line is bound to fail unless there already is some /dev/md? ... which gets assembled later | 04:12 |
| gnarface | yea, labels might be a smarter choice | 04:12 |
| cehteh | https://paste.debian.net/1203864/ puzzleed about whats partition whats filesystem and what the raid uses | 04:13 |
| gnarface | well they're all partitions | 04:15 |
| cehteh | see that the UUID of ARRAY md0 appears 3 times below :D | 04:15 |
| gnarface | the important part is that the ones starting with nvme0* and sd* are physically existing partitions | 04:15 |
| gnarface | the others are virtual of some sort | 04:15 |
| cehteh | can mdadm.conf use labels? | 04:15 |
| gnarface | don't know for sure but i would assume so | 04:16 |
| gnarface | i mean if it can use UUIDs... | 04:16 |
| cehteh | stil lthings look correct | 04:16 |
| gnarface | i think the uuid's are NOT supposed to repeat | 04:17 |
| gnarface | that seems to be in error | 04:17 |
| gnarface | afaik the whole point of a uuid is that it's globally and temporally unique unless something has gone very wrong | 04:17 |
| gnarface | it's right there in the name | 04:17 |
| cehteh | i tihnk thats the UUID of the raid | 04:17 |
| rrq | cehteh: are you up to adding a # to line 53? | 04:17 |
| gnarface | but, that said... you can hose your UUID algorithm with bad clock times in theory... though a collision doesn't seem likely anyway | 04:17 |
| cehteh | well i am too tired now continue tomorrow | 04:18 |
| cehteh | and yes the uuids repeats for the other array too . .thats the shared uuid for the entire array | 04:18 |
| cehteh | lemme check on this computer but i tihnk thats correct | 04:19 |
| cehteh | yes its common to repeast | 04:19 |
| gnarface | if it is common for the uuid of the raid to match one of the partitions in it, then maybe it's just got the wrong one somehow? | 04:20 |
| cehteh | https://paste.debian.net/1203865/ <<< i have more fancs working setups :D | 04:20 |
| gnarface | i admit last time i did this i was not using uuid's either | 04:20 |
| Xenguy | It's official, you guys are off the hook, keep on rockin' in the free world | 04:23 |
| cehteh | rrq: line53 on what? /usr/share/initramfs-tools/scripts/local-block/mdadm ? | 04:24 |
| cehteh | rrq: that looks like thats the symptom of the error but that it doesnt recognize the arrays should be earlier | 04:25 |
| cehteh | grr .. i suspect udev | 04:26 |
| cehteh | what is eudev vs udev from backports? | 04:29 |
| gnarface | eudev is devuan's fork | 04:29 |
| gnarface | it is supposed to be a drop-in replacement | 04:30 |
| cehteh | 'supposed to be' :) | 04:30 |
| cehteh | well i am just thinking if i can install the other udev to cross check | 04:30 |
| gnarface | i had different but similar problems with both of them on ceres | 04:30 |
| gnarface | and the issues were isolated to only certain devices | 04:31 |
| gnarface | (mostly usb stuff) | 04:31 |
| fsmithred | udev in beowulf-backports probably should not be there. | 04:31 |
| fsmithred | We have a udev package in beowulf, chimaera and ceres, and they all have +devuan in the version. The one in bpo does not. | 04:32 |
| fsmithred | also it matches debian version numbers: 247x instead of 1:3.2x | 04:33 |
| gnarface | hmm, an upgraded system would leave the debian udev in place though rather than replacing it with eudev, wouldn't it? | 04:33 |
| gnarface | i thought i remember having to switch mine manually | 04:34 |
| fsmithred | no, we added the epoch to replace the debian one | 04:34 |
| fsmithred | 1: makes it later than 247 or anything else | 04:34 |
| fsmithred | except 2: | 04:34 |
| cehteh | ahh .. i had eudev an udev mixed installed, removing udev remains | 04:35 |
| cehteh | mhm these dont share names/collide strange .. | 04:37 |
| fsmithred | I think we had a dummy udev package, maybe in ascii | 04:38 |
| rrq | cehteh: would be worth a try with the commenting (#) + update initrd + reboot ..? | 04:38 |
| cehteh | noted but i am just fooling around currently i am too tired trying that tomorrow | 04:39 |
| * rrq time for lunch here :) | 04:39 | |
| cehteh | that also looks like it fails but is not the cause | 04:39 |
| cehteh | 4:39am here | 04:39 |
| cehteh | and purging the old udev remains didnt fixed it | 04:40 |
| cehteh | done for today | 04:40 |
| user_ | reading backlog: cehteh: echo $'\x2d' -> - | 08:25 |
| user_ | this is in bash, in sh you may have to use printf | 08:26 |
| user_ | echo -e "\x2d" should work in ash too | 08:26 |
| user_ | iow printf --help can be coded: `echo -e "printf \x2d\x2dhelp"` | 08:27 |
| user_ | `` executes the string returned by echo | 08:27 |
| user___ | I know I am late to the party, but better later than never. | 08:29 |
| user___ | cehteh: your can't-enter-hyphen problem is a worthy support issue, you are a paying customer, should get imediate help on that from the resident admin even at night. It's probably something simple anyway. | 08:30 |
| user___ | https://docs.hetzner.com/robot/dedicated-server/maintainance/kvm-console/ wow 8.5 EUR/hr above 3 hours?! | 08:31 |
| user___ | Also, full instructions on how to support your keyboard in there, it is done graphically, in the kvm ui, not using loadkeys | 08:34 |
| user___ | I must have come in too early for folks on here usually :) | 08:56 |
| user___ | 8.40 EUR/3hrs of hw kvm use is extortion. After 9 hrs of use you can buy a decent KVM of your own for the money, including shipping to Hetzner... | 08:59 |
| user___ | Alas, I see the gaming / HDtv crowd have hijacked the KVM term again, it's impossible to find a PC-kvm by kvm alone. This after the kernel people hijacked kvm for kernel. Guys, there are 26 letters in the alphabet, spread out a bit. | 09:02 |
| user___ | </rant> | 09:02 |
| user___ | 'kvm over ip price' search works | 09:06 |
| user___ | Ok, I retract the above "after 3 hours you can buy...", the ip-kms got a lot more expensive lately. | 09:11 |
| user___ | On topic: is there a specific devuan image "tested" in rack / remote vm / data center mode which is preferred? One with minimal installation, no gui loaded, and accent on manual network configuration? | 09:14 |
| aloo_shu | I'm hitting a wall, I got a small devuan ascii rootfs in a chroot under android (4.2.2, kernel 3.3.0), all nice, /dev /dev/pts /proc and /sys are mounted over. When I'm trying ping 8.8.8.8, I'm getting socket: Permission denied | 09:14 |
| user___ | Is the android rooted? | 09:15 |
| user___ | Also, which ping -> post output | 09:15 |
| aloo_shu | also mounting over android's /system/bin and using androids ping from devuan's bash, works | 09:15 |
| aloo_shu | the android is rooted, I'm root in- and outside the chroot | 09:15 |
| user___ | Okay. | 09:16 |
| aloo_shu | /bin/ping (in devuan) | 09:16 |
| user___ | dns queries use port 53 which is privileged, is it possible selinux or something else security related thinks that your chrooted system should not have access to root moe sockets (<1000) ? | 09:17 |
| user___ | I mean ping uses port 0 not dns, sigh | 09:17 |
| user___ | Also, this is really not a #devuan question, you may have more luck in #android-devel ? | 09:18 |
| schillingklaus | will it take again a year for devuan to move to the next major version after debian does? | 09:18 |
| aloo_shu | I *think* android 4.2.2 isn't yet that heavy on selinux - how could i check | 09:18 |
| user___ | You are right, but you should ask in $above, not here | 09:18 |
| * user___ notes even kvm-over-ip is hijacked by the same hdmi-extender crowd, they call it "over ip" because it uses RJ45 connected CAT6 cable ARGH | 09:20 | |
| user___ | The real pc-kvm-over-ip boxes got a lot more expensive to support crazy video modes on video cards. Back in the days when I dealt with these things, only SVGA needed to be supported, dead easy with relatively simple cheap chips. | 09:25 |
| user___ | bbl | 09:32 |
| aloo_shu | anybody else? android security limiting the chroot env is pretty ruled out, I put a busybox binary into my chroot, busybox ping is working | 09:42 |
| gnarface | something to do with failed ipv6 routing? just guessing | 09:43 |
| aloo_shu | I'm specifically pinging 8.8.8.8 | 09:44 |
| gnarface | could just be crappy security, badly implemented, and your clever ruse completely fooled it | 09:47 |
| onefang | Recent Android version have toybox installed by default, but you said this was an old Android? | 09:49 |
| schillingklaus | do recent androids still use mksh? | 09:49 |
| aloo_shu | I rather think something is missing that devuan is expecting, maybe 3.3.0. kernel too old? using pung verbosely, I see it's trying to open a socket, then a raw socket after that fails, and is getting permission denied/protocol not supported errors | 09:50 |
| onefang | But as was said before, take Android support stuff to an Android support channel. This is a Devuan support channel. | 09:50 |
| aloo_shu | dunno, this android is all-but-recent (4.2.2) | 09:50 |
| aloo_shu | I am trying to find out how devuan networking tools are trying to open sockets, and what they require to work | 09:51 |
| schillingklaus | if he needs to write portable scripts for both devuan and the native android system, it is better to know the available tools on both | 09:51 |
| gnarface | afaik they should only need device permission in /dev | 09:51 |
| gnarface | i assume you have that or the busybox one wouldn't work either | 09:52 |
| aloo_shu | ^ | 09:52 |
| gnarface | kernel api seems plausible too | 09:52 |
| onefang | The author of toybox uses Devuan, and used to be the maintainer of busybox. | 09:52 |
| schillingklaus | Rob Landley | 09:53 |
| onefang | Yep. | 09:53 |
| schillingklaus | but he is also active in the musl world | 09:53 |
| schillingklaus | for musl is good for static compilation, which is an advantage for many intended usages of toybox | 09:54 |
| aloo_shu | I have had a debian wheezy chroot system with working networking on the same android before, what I cannot say, if this is due to the proximity in kernel version, or the fact that I used an app to configure the chroot that time, while I'm trying manually this time | 09:56 |
| aplainzetakind | I installed from netinst, but had wifi troubles (from the network) so I finished without downloading anything extra and ended up with the minimal system. | 13:36 |
| aplainzetakind | Is there a clean way to continue from a boot into this state or am I better off just reinstalling? | 13:37 |
| fsmithred | aplainzetakind, you're in the right spot to add what you want. If you want the full desktop, install task-xfce-desktop or similar. | 13:39 |
| fsmithred | otherwise, start installing the individual things you want. | 13:39 |
| user___ | does he not need to make wifi work 1st? | 13:40 |
| fsmithred | oh, maybe | 13:40 |
| aplainzetakind | Yeah I don't have wifi. | 13:40 |
| fsmithred | if there's no wire | 13:40 |
| * user___ cheats and uses an openwrt flashed older router as wifi ap in client mode (!), to supply RJ-45 net to balky boxes while installing | 13:40 | |
| fsmithred | nice | 13:40 |
| user___ | klunky | 13:41 |
| aplainzetakind | Nor ethernet. | 13:41 |
| user___ | aplainzetakind: do you have a rj-45 at all? and wired ethernet nearby? | 13:41 |
| fsmithred | other options would be to install from the desktop dvd or server cd or one of the live isos. Any of those can give you a usable system without needing a netork connection. | 13:41 |
| fsmithred | and the live isos have wireless firmware already installed | 13:42 |
| user___ | aplainzetakind: w/o wired net and wifi your usual option at this point is to d/l a bunch of relevant .deb packages to usb stick and mount that and use | 13:42 |
| fsmithred | (except for a couple of tricky broadcoms) | 13:42 |
| user___ | and what fsmithred said | 13:42 |
| aplainzetakind | user___: No cable nor ethernet connectivity. | 13:42 |
| aplainzetakind | I mean it's not too much trouble to do over with -desktop. | 13:42 |
| user___ | aplainzetakind: ok so reinstall as fsmithred said or usb stick transfer what is needed | 13:42 |
| user___ | wifi drivers are always a pita in linux, because of the non free allergy. | 13:43 |
| fsmithred | aplainzetakind, were you able to pull anything from the network, or did wifi not work at all? | 13:43 |
| user___ | meaning, people hating to package nonfree drivers. | 13:43 |
| fsmithred | the worst are the broadcoms that require a network connection to get the firmware that allows you to have a network connection. | 13:44 |
| user___ | yes. And they're not the only ones. | 13:45 |
| * user___ also has an ancient ZD1211 iirc based usb stick wifi for stubborn cases, that has a kernel driver. | 13:45 | |
| aplainzetakind | I'm at the uni, there is a PEAP wifi which needs a username/pass pair, so that didn't seem to be an option and I logged into the Guest wifi, but that has the initial confirmation page nonsense. | 13:45 |
| aplainzetakind | But, I will also need non-free drivers. | 13:46 |
| aplainzetakind | I think. | 13:46 |
| user___ | aplainzetakind: oh, ok, do you have data on the phone? Open a connection sharing hotspot? | 13:46 |
| aplainzetakind | user___: I can do that but I don't have an abundance of data. How much would it consume? | 13:47 |
| user___ | Doesn't max 4GB I assume. A dvd worth. Likely much less. | 13:47 |
| aplainzetakind | I can create the usb from my other machine that's no problem. I think I'll be fine. | 13:48 |
| aplainzetakind | Does the -desktop installer offer the proprietary driver? | 13:48 |
| user___ | aplainzetakind: do you see ssid's in the output of iw scan ? | 13:48 |
| user___ | now as is? | 13:48 |
| fsmithred | most wireless firmware is in the iso. You can almost always ignore that first window that asks you to load firmware from usb. | 13:48 |
| aplainzetakind | user___: With this installed minimal state? ifconfig doesn't show wifi. | 13:49 |
| aplainzetakind | Nor ethernet actually, only lo. | 13:49 |
| user___ | what about dmesg? | 13:49 |
| user___ | interface needs to be up to show. ifconfig -a might help | 13:49 |
| fsmithred | brb, reboot router | 13:50 |
| user___ | also driver loaded for wifi and eth | 13:50 |
| aplainzetakind | Hmm, dmesg says some iwlwifi has been loaded. | 13:50 |
| user___ | nice, which one. | 13:50 |
| aplainzetakind | 8265-36 | 13:50 |
| aplainzetakind | So why don't I have tha interface on ifconfig? | 13:51 |
| user___ | and does it then list a device? Show the dmesg lines for this | 13:51 |
| aplainzetakind | Ffff I forgot -a | 13:51 |
| user___ | Even ifconfig -a does not show it if the driver is not loaded | 13:51 |
| aplainzetakind | It does though. | 13:52 |
| user___ | Do you see a wifi0: in dmesg? | 13:52 |
| aplainzetakind | I have wlan0 | 13:52 |
| user___ | Nice. | 13:52 |
| user___ | Now try: ifconfig wlan0 up | 13:52 |
| user___ | iw scan wlan0 | 13:52 |
| user___ | sorry iw wlan0 scan | 13:52 |
| aplainzetakind | wlan0 is down | 13:53 |
| aplainzetakind | ifup wlan0 says no such interface. | 13:53 |
| user___ | it needs a ssid to be brought up as sta | 13:53 |
| user___ | do you know how to configure manually? | 13:53 |
| aplainzetakind | Not really. I probably did it once. | 13:54 |
| user___ | https://linuxconfig.org/connect-to-wifi-from-the-linux-command-line | 13:54 |
| user___ | do you have wpa_cli ? | 13:54 |
| aplainzetakind | Yes. | 13:55 |
| user___ | then follow the guide in the link | 13:55 |
| aplainzetakind | wpa_supplicant is not up though. | 13:56 |
| user___ | start it from /etc/init.d | 13:56 |
| user___ | no, manually | 13:57 |
| aplainzetakind | user___: Eh. What was the command to issue? | 14:01 |
| fsmithred_ | wpa_cli? | 14:03 |
| aplainzetakind | I always get tied up with wpa_supplicant | 14:03 |
| fsmithred_ | he said to start something from /etc/init.d but wpa_supplicant is not there | 14:04 |
| aplainzetakind | -B -D nl80211 -i wlan0 or something I guess. | 14:04 |
| fsmithred_ | I don't do it manually, so I have no idea. | 14:05 |
| fsmithred_ | oh shit. I already mentioned there's no init script, but there are service files | 14:05 |
| aplainzetakind | Yeah there should be files for interfaces I think. | 14:05 |
| aplainzetakind | I don't have them, but I think I should create one. | 14:06 |
| fsmithred_ | yeah, you can put a stanza in for wlan0 | 14:06 |
| Guest4 | hello friends | 14:09 |
| Guest4 | I was away from freenode for a while. | 14:09 |
| Guest4 | I came back to an empty house | 14:10 |
| Weeezy | why are we here? | 14:10 |
| fsmithred_ | you really don't know what happened? | 14:11 |
| Weeezy | No I don't, sorry. | 14:11 |
| fsmithred_ | google for freenode and andrew lee | 14:11 |
| Weeezy | ok. thanks. | 14:12 |
| fsmithred_ | hostile takeover | 14:12 |
| Weeezy | my registered login was deleted on freenode | 14:12 |
| fsmithred_ | oh yeah. They all were. | 14:13 |
| Weeezy | why this network tho? | 14:13 |
| fsmithred_ | this is the fork. Almost all the staff left to fork this. | 14:13 |
| Weeezy | ok. | 14:14 |
| fsmithred_ | and many channels moved with the staff. (community) | 14:14 |
| user___ | aplainzetakind: if you don't have a config for supplicant, generate one using: `wpa_passphrase networkname password > /etc/wpa_supplicant/wpa_supplicant.conf` | 14:16 |
| user___ | aplainzetakind: this is in the manual setup link I posted above | 14:16 |
| user___ | aplainzetakind: then start wpa_supplicant with that config | 14:16 |
| user___ | you still need to say which interface with -i and daemon -B | 14:17 |
| aplainzetakind | Ah alright. Thanks. | 14:17 |
| user___ | iw wlan0 scan; should output the ssid's it sees. Try that 1st | 14:17 |
| user___ | if you don't see ssid's there is more driver work ahead. But Intel drivers should work. | 14:18 |
| user___ | I.e. no more driver setup needed. | 14:18 |
| aplainzetakind | I will write a dummy config and manually edit it to the PEAP network stanza from the other computer. | 14:18 |
| user___ | yes | 14:18 |
| user___ | good plan | 14:18 |
| user___ | and it's: iw dev wlan0 scan | 14:20 |
| user___ | https://wiki.archlinux.org/title/Network_configuration/Wireless#iw_and_wireless_tools_comparison good resource, as usual | 14:20 |
| Weeezy | was or is this Lee guy the one using bots to spam other networks? My quakenet channels were all being spammed with libera.chat propaganda | 14:22 |
| aplainzetakind | Isn't the way to specify the file -I ? | 14:22 |
| user___ | Weeezy: that was a false flag operation done by disgruntled freenode evil geniuses | 14:23 |
| user___ | aplainzetakind: -c | 14:23 |
| Weeezy | that's what I was thinking | 14:23 |
| user___ | aplainzetakind: -i $interface -c $config_file -B | 14:23 |
| Weeezy | thanks. | 14:23 |
| user___ | Weeezy: freenode evil geniuse means NOT real freenode staff or users | 14:23 |
| user___ | +s | 14:24 |
| aplainzetakind | user___: Alright wifi is up. | 14:24 |
| user___ | nice. And are you connected? | 14:24 |
| user___ | ping? | 14:24 |
| aplainzetakind | No, for some reason. | 14:24 |
| aplainzetakind | Let me try my phone's hotspot instead. | 14:25 |
| user___ | right, edit /etc/resolv.conf | 14:25 |
| aplainzetakind | That PEAP thing may be hairy. | 14:25 |
| user___ | do you have dhclient running or not? | 14:25 |
| aplainzetakind | user___: | 14:25 |
| user___ | do you have an IP | 14:25 |
| aplainzetakind | No. | 14:25 |
| user___ | ... | 14:25 |
| aplainzetakind | ip r is silent. | 14:25 |
| user___ | so maybe dhclient wlan0 ... | 14:25 |
| aplainzetakind | Cool, can ping. | 14:26 |
| aplainzetakind | Thanks. | 14:26 |
| user___ | then find out the ns address (usually the router if nat'd), and edit /etc/resolv.conf | 14:26 |
| aplainzetakind | Should I hardcode that at this moment? | 14:27 |
| aplainzetakind | This is just a momentary network. | 14:27 |
| aplainzetakind | Aren't these manual settings to be superseded when I install more stuff? | 14:27 |
| user___ | yes, wicd or networkmanager will do all this for you | 14:28 |
| user___ | OR you can write /etc/init.d/wlan0net and have it linked in your runlevel by hand | 14:28 |
| user___ | OR put it all in /etc/rc.local until you finish installing | 14:29 |
| user___ | OR write all to a shell command file like /root/wlan0net.sh and source that as root after rebooting | 14:29 |
| user___ | :) | 14:29 |
| user___ | OR install systemd and write a unit file for it, and make it work for you (see you in a few weeks if you go this way) | 14:29 |
| aplainzetakind | Haha, no. | 14:30 |
| aplainzetakind | I run bare wpa_supplicant via wpa_gui on my other computer. | 14:30 |
| aplainzetakind | But I want to try out connman. | 14:30 |
| user___ | So, in a nutshell, what you did so far: wpa_supplicant -i wlan0 -c /etc/wpa_supplicant/wpa_supplicant.conf -B; dhclient wlan0; cat my_resolv_conf >/etc/resolv.conf | 14:31 |
| aplainzetakind | I didn't do that last bit. | 14:32 |
| user___ | What's the output of host google.com | 14:32 |
| aplainzetakind | I don't have host. | 14:32 |
| aplainzetakind | Let me get it. | 14:32 |
| user___ | yes | 14:32 |
| aplainzetakind | Hmm I don't have repos configured. Only looking at cdrom:// stuff. | 14:33 |
| aplainzetakind | Hold on. | 14:33 |
| user___ | aplainzetakind: you need dns before you can use named hosts... | 14:34 |
| user___ | aplainzetakind: try this (I hate myself): cat "8.8.8.8" >/etc/resolv.conf | 14:34 |
| user___ | in theory dhclient should configure resolv.conf so you can skip that | 14:34 |
| user___ | you can try ping ping.com | 14:34 |
| user___ | It will stall but show the resolved ip | 14:35 |
| user___ | kill it with ^C | 14:35 |
| user___ | bing.be hates the fact they got that host name >:-) | 14:35 |
| aplainzetakind | I can ping google.com I have name resolution. | 14:35 |
| user___ | Ah it's no longer owned by ping.be, it's someone in Phoenix, AZ | 14:36 |
| user___ | aplainzetakind: okay then go on and configure sources.list and continue installation | 14:36 |
| user___ | aplainzetakind: imo, etch the steps to get networking going on linux into your dna. What you did today :) | 14:37 |
| user___ | Note pinging external hosts may not work in some lan/walled garden situations, but name resolution should work, and ping prints that in the 1st line of output even if there is no echo reply. | 14:38 |
| user___ | and it was: echo "nameserver 8.8.8.8" >/etc/resolv.conf above. Sigh, it's been a while since I needed this. | 14:40 |
| aplainzetakind | Alright, sources.list good, apt-get update'd. | 14:41 |
| aplainzetakind | How do I "continue with the installation"? | 14:42 |
| aplainzetakind | user___: I want to eventually startx into xmonad with no DE or anything. | 14:46 |
| user___ | Well, install what you want to run. I don't know about xmonad. | 14:46 |
| user___ | I use aptitude usually to search for and install packages | 14:47 |
| user___ | Usually one needs some sort of login manager. Especially on a public machine. Like slim | 14:48 |
| aplainzetakind | This is a laptop. | 14:48 |
| user___ | Is it your or shared? | 14:49 |
| user___ | You probably want a login there anyway. | 14:49 |
| user___ | slim can do this for you | 14:49 |
| user___ | Unless xmonad handles xsession logins | 14:49 |
| * user___ installs xnee | 14:49 | |
| aplainzetakind | Mine. | 14:50 |
| aplainzetakind | I think I'll be fine with plain startx. | 14:50 |
| user___ | okay. | 14:50 |
| aplainzetakind | I'll get some screen locker. | 14:50 |
| user___ | your machine, your rules | 14:50 |
| aplainzetakind | But, with openrc, is installing connman and doing rc-update add connmand sufficient? | 14:52 |
| fsmithred_ | aplainzetakind, I haven't used connman with openrc, but I would expect that just installing connman is enough. Openrc uses the sysvinit scripts. | 15:07 |
| fsmithred_ | so does runit | 15:07 |
| aplainzetakind | fsmithred_: Turned out it is. | 15:08 |
| fsmithred_ | I know there are runit scripts for connman that work, but I don't know about openrc. | 15:09 |
| fsmithred_ | probably need to look at gentoo for examples | 15:09 |
| Weeezy | this is not a huge deal but I'm wondering why I can't get devuan to report v 3.1 | 15:11 |
| Weeezy | I've tried all the commands I know but still reports v 3 | 15:12 |
| fsmithred_ | Weeezy, because we didn't update base-files to say 3.1 | 15:12 |
| user___ | I see we don't have recapp in devuan? com.github.amikha1lov.RecApp ? | 15:12 |
| fsmithred_ | if it said 3.1 and you did update/upgrade, the 3.1 would no longer reflect the true state of your system. | 15:13 |
| fsmithred_ | not 3.2, but newer than 3.1 | 15:13 |
| fsmithred_ | the point releases are releases of new isos | 15:13 |
| fsmithred_ | the number reflects what's in the iso, not what's in your installed system | 15:14 |
| user___ | we also seem to not have xrebind from gnu | 15:14 |
| fsmithred_ | does debian have those packages? | 15:14 |
| user___ | Looking. gnu certainly, recapp is in fedora and is open source. | 15:15 |
| fsmithred_ | we don't take packages from fedora, we take them from debian | 15:15 |
| fsmithred_ | and those two are not in debian that I can see | 15:16 |
| user___ | xrebind seems to be 2006 vintage... | 15:16 |
| user___ | xrebind allows keys to be bound to macros in X11, which can be useful, since macros will also issue mouse moves, etc. | 15:17 |
| fsmithred_ | is it older than xvidtune? | 15:17 |
| user___ | I don't know. Trying to build | 15:18 |
| user___ | ../include/xrebind.h:34:10: fatal error: X11/extensions/XTest.h: No such file or directory | 15:18 |
| user___ | libx11-dev is installed | 15:19 |
| user___ | libxext-dev too | 15:20 |
| fsmithred_ | libxtst-dev: /usr/include/X11/extensions/XTest.h | 15:21 |
| user___ | don't have that | 15:21 |
| user___ | ok, found it, wrong re search | 15:22 |
| user___ | +libxosd-dev | 15:23 |
| user___ | xrebind built on beowulf with minor erros | 15:24 |
| user___ | *errors | 15:24 |
| fsmithred_ | time to go. bbl. | 15:27 |
| user___ | works fine with examples | 15:27 |
| user___ | thanks, bye | 15:27 |
| user___ | grr gnu makefile does not install info and man pages | 15:29 |
| user___ | it is not debianized | 15:30 |
| enyc | user___: what do you mean gnu-makefile -- the "Makefile" included in source-package of a particular GNU official project ??? | 15:31 |
| user___ | The output of ./configure in package xrebind-0.6.tar.gz | 15:42 |
| user___ | I hacked a manpage from the output of xrebind --help if anyone needs it say beep | 15:43 |
| user___ | Should bluez-tools work without pulse? | 15:54 |
| GyrosGeier | non-audio bluetooth stuff should always work without pulse | 15:56 |
| GyrosGeier | audio bluetooth stuff is old | 15:56 |
| GyrosGeier | and pulse have a userspace implementation of the protocols | 15:57 |
| GyrosGeier | so if you want headsets to work without pulse you need a kernel module | 15:57 |
| cehteh | user___: unfortunally these docs seem to be out of date, the kvm console opens here as webapp not java (which sucked years ago, but i may try to fall back to that if possible) and there was no way to define the keyboardlayout in the webapp | 16:26 |
| user___ | Hmm | 16:43 |
| user___ | cehteh: open an urgent support ticket? | 16:43 |
| user___ | cehteh: echo -e "\x2d" works aanyway as a hack | 16:44 |
| cehteh | nit when you cant enter \ either and | is missing too :D | 16:44 |
| cehteh | i opened a support ticket already | 16:44 |
| user___ | ouch | 16:44 |
| cehteh | almost any symbols are missing or out of place, and its not just german vs us layout because - gave ß | 16:45 |
| user___ | On a US 104 key keyboard - is the key to the right of '0' | 16:45 |
| user___ | I assume you tried all with and without shift | 16:46 |
| user___ | Do you get underline _ ? The - is normally above it or under it (unshifted) | 16:46 |
| cehteh | yes o know | 16:46 |
| cehteh | on german keyboiard - is rightmost lowest row, besides shift | 16:47 |
| cehteh | and the ß is right of 0 | 16:47 |
| cehteh | ahh i need coffee | 16:47 |
| cehteh | äöü worked surprisingly | 16:47 |
| cehteh | so its somehow totally messed up | 16:48 |
| cehteh | () where like on us | 16:48 |
| user___ | I know where it should be on .de and .us | 16:48 |
| cehteh | and no " either | 16:48 |
| user___ | wow | 16:48 |
| user___ | Did you try another browser? | 16:48 |
| cehteh | not yet i gone to sleep | 16:48 |
| cehteh | can try chromium leter | 16:48 |
| user___ | Try another browser. | 16:49 |
| user___ | What did you try until now? ff? | 16:49 |
| cehteh | yes | 16:49 |
| cehteh | well its not ultra-urgend, no business in danger | 16:49 |
| user___ | ff has some problems with newer js / css. No doubt deliberate (/me shakes fist at google) | 16:49 |
| user___ | cehteh: 8.4 EUR + VAT / 3 hrs of KVM is a strong incentive to get it going fast. | 16:50 |
| cehteh | first 3 hrs are free and you can reopen the ticket, they only charge because they dont want peopel to hoard the consoles and have them continously connected | 16:52 |
| cehteh | hetzner support is pretty ok, i know the drill, yesterday i asked for 2 hours (just precaution, didnt know about the problem yet) but stated "its no urgend, but can you leave the console connected as long as possible over night" .. and i had the console the whole night | 16:54 |
| user___ | look at the next bill... | 16:54 |
| cehteh | will do but i had that before, usually its no problem | 16:58 |
| cehteh | anyway whiel eating now, i need to think about how to find and fix this problem | 16:59 |
| cehteh | can i turn initrd into some kind of verbose mode? enable set -x from kernel command line? | 17:03 |
| user___ | Not that I know of | 17:03 |
| user___ | set -x inside initrd init script might work | 17:03 |
| user___ | i.e. /init inside initrd | 17:04 |
| cehteh | yes but iirc (from long ago) i remmeber there was some debugging help, i just look that up now | 17:04 |
| user___ | er /linuxrc | 17:05 |
| cehteh | /linuxrc ? | 17:07 |
| user___ | /linuxrc or /sbin/init in the initrd is what is run by the kernel. | 17:07 |
| cehteh | ah | 17:08 |
| user___ | it is a sh script (not bash!) | 17:08 |
| user___ | set -x should work | 17:08 |
| cehteh | ok got the console now .. finishing eating and then go | 17:09 |
| user___ | remove bread crumbs from keyboard to make '-' key work... | 17:09 |
| cehteh | no bread, no muesli :) .. was saussage and carbagge salad :D | 17:10 |
| user___ | careful with npm updates on devuan, you'll get scary warnings https://www.theregister.com/2021/07/09/npm_audit_security/ | 17:12 |
| cehteh | they didnt answer me for the keyboars stuff :/ trying chromium now | 17:13 |
| user___ | tell us how it goes | 17:13 |
| cehteh | no does not work | 17:13 |
| user___ | hmm, tell support you also tried chromium. | 17:13 |
| user___ | cehteh: copy paste into web terminal works? | 17:15 |
| cehteh | nope | 17:15 |
| user___ | wow, report that too to support | 17:15 |
| user___ | basically you have non working kvm-via-ip now | 17:16 |
| cehteh | its just a vga screengrabber sending a video stream | 17:16 |
| user___ | we are interested in the other direction | 17:16 |
| cehteh | yeah | 17:16 |
| user___ | maybe there's a "security" filter they put in to prevent miscreants who gain access to cause trouble? | 17:17 |
| cehteh | i tihnk they just migrated/updated from java to webapp and its broken | 17:17 |
| user___ | sounds like it, so far. Nothing this broken would ever be deployed at a vh provider. | 17:17 |
| user___ | sounds like a filter. | 17:17 |
| cehteh | ir maybe i need to install java here to fall back to java | 17:18 |
| user___ | you can try that but you should wait for their answer. | 17:18 |
| user___ | Seriously, web shells in js work perfectly, like shellinabox | 17:18 |
| user___ | It is trivial to mod it to show a video stream instead of a real terminal. | 17:18 |
| cehteh | what is even needed to enable java from browser? havent done that in ages | 17:19 |
| user___ | Oh you can forget about that, every warning will light up. | 17:19 |
| user___ | Google is not on good terms with Oracle | 17:20 |
| GyrosGeier | nah, that's just banter between friends | 17:20 |
| user___ | GyrosGeier: who occasionally whack each other's head with golf clubs. | 17:20 |
| user___ | https://java.com/en/download/help/enable_browser.html difficulties? With the database? Of a database provider? Hah. | 17:21 |
| cehteh | mhm i suppose changing keyboard layout in the bios wont work, but i give it a try | 17:21 |
| GyrosGeier | well, they might disagree on licensing, but they still won't poach each others' employees. | 17:21 |
| cehteh | searching if that bios has some other remote admin stuf like ipmi or so | 17:21 |
| user___ | GyrosGeier: of course not, that would push salaries up. | 17:22 |
| user___ | cehteh: that may be even harder to connect to. | 17:22 |
| cehteh | possibly | 17:23 |
| user___ | cehteh: install jre and the plugin, for firefox. Then enable plugin, exit and restart the browser. In theory. | 17:25 |
| cehteh | ok bios keyboard change didnt help (as expected) | 17:29 |
| user___ | cehteh: I am now 90% sure it's a "security" feature. | 17:33 |
| user___ | Ask support if they mistakenly enabled some kind of security filter on the kvm | 17:33 |
| cehteh | i doubt, but i opened a ticket | 17:33 |
| user___ | Also, can you copy/wget a url from the initrd? | 17:33 |
| cehteh | what? | 17:34 |
| user___ | You can pack single character files with what you need and unpack them there, then `cat $bar` etc | 17:34 |
| user___ | $pipe | 17:34 |
| cehteh | https://pbs.twimg.com/media/E5z2El5WUAYRgI1?format=jpg&name=4096x4096 << thats a screenshot of the console | 17:35 |
| user___ | this is a crazy situation imo. Support should help you asap. | 17:35 |
| cehteh | yes | 17:35 |
| user___ | what does the menu tab on the left do? | 17:35 |
| cehteh | no keyboard layout switcher there | 17:36 |
| user___ | busybox should have wget | 17:36 |
| cehteh | some menus i check them all | 17:36 |
| cehteh | initrd doesnt have network running | 17:36 |
| cehteh | i may setup dropbear in initrd but thats famous for other problems | 17:36 |
| user___ | that is a crap kvm, using letterbox mode due to low resolution. Maybe ask for another kvm? I assume the nonfree ones are not buggy, else there would be hell to pay. | 17:36 |
| cehteh | i cna fullscreen it :D | 17:37 |
| user___ | sigh | 17:37 |
| cehteh | but 4k monitor here | 17:37 |
| cehteh | anyway its just vga screengrabber, no more no less | 17:37 |
| cehteh | and i want to focus on the problem with booting now rather debugging remote consoles | 17:37 |
| cehteh | i am in the rescue system again and can hack the initrd | 17:38 |
| cehteh | the rescue system start the raid arrays even without knowing about them | 17:39 |
| user___ | did you do what fsmithred_ said? About initrds from rescue/stock system? | 17:41 |
| cehteh | no lemme scroll up | 17:41 |
| cehteh | huh what he saied a lot :D | 17:42 |
| cehteh | ah initrd parses the commandline and has a debug mode | 17:44 |
| Harzilein | cehteh: look into that hyper.m4 thing that used to come with consoleutils (?, kbd-utils?) | 17:47 |
| cehteh | wtf what? | 17:47 |
| Harzilein | cehteh: then you should be able to enter both hex and decimal ascii codes for the keys | 17:47 |
| Harzilein | cehteh: had it when i had a 85 key keyboard | 17:47 |
| cehteh | i am in initrd | 17:47 |
| Harzilein | but there's a loadkeys there? | 17:47 |
| cehteh | tell me how to install a keyboard layout in initrd AND provide me one that fixes this borked layout | 17:48 |
| cehteh | thats prolly an afternoon work alone | 17:48 |
| Harzilein | kk | 17:48 |
| Harzilein | yeah, sorry for digressing | 17:48 |
| cehteh | anyway i am now looking at the initrd log :) | 17:49 |
| cehteh | all i see is that it somehow doesnt find device by UUID, but the erorr must be before that it doesnt start the raid | 18:05 |
| cehteh | wow i found the - !!!! | 18:09 |
| cehteh | ok problem: mdadm --detail --scan should list all discovered raids (config) | 18:12 |
| cehteh | but doesnt do any | 18:12 |
| user___ | where was - | 18:13 |
| cehteh | on the # | 18:13 |
| user___ | O.o | 18:13 |
| user___ | You did say you tried all keys ... did you find | too? | 18:14 |
| cehteh | overlooked that before, did i say that this console has half a second delay | 18:14 |
| cehteh | not yet :D | 18:14 |
| user___ | Eww | 18:14 |
| cehteh | and some keys generates multi key event incl delete / backspace chars | 18:14 |
| user___ | just cat <<EOF >/dev/null and type all keys, then EOF | 18:15 |
| user___ | ok, that is fun. Nothing from support? | 18:15 |
| cehteh | then i dont know what relates to what key | 18:15 |
| user___ | (now you know why they let you have this kvm all night with no payment) | 18:15 |
| cehteh | waiting, ticket received | 18:15 |
| cehteh | anyway with - i can go on | 18:15 |
| user___ | find the pipe and >< too, soon. | 18:15 |
| cehteh | i see that the kernel/mdamd somehow doesnt find any raid member | 18:16 |
| cehteh | mdadm v4.1 is correct? | 18:16 |
| cehteh | wasnt there a command to make the kernel to rescan/rediscover all block devices? | 18:18 |
| user___ | 4.1 is what I have on Beowulf | 18:18 |
| cehteh | well blkid does list them | 18:20 |
| user___ | cehteh: https://serverfault.com/questions/36038/reread-partition-table-without-rebooting | 18:20 |
| user___ | speaking of block*, you do hatev the block* tools on the initrd? | 18:21 |
| user___ | blockdev blkid etc | 18:21 |
| user___ | blockdev --rereadpt /dev/sdX ; is rumored to work | 18:22 |
| cehteh | partprobe is available worked .. but mdadm / raid doesnt pick that up wtf | 18:22 |
| user___ | the page suggests 'systemctl restart systemd-udevd.service' <evil grin> | 18:23 |
| user___ | so, ahem, is udevd running? eudevd perhaps? | 18:24 |
| cehteh | systemd in initrd :) | 18:24 |
| user___ | yup. | 18:24 |
| user___ | Hey, it *is* the kernel already... | 18:24 |
| user___ | that is from my link above. The scary part: comment: "The top accepted answer is incomplete: in the modern systemd world, THIS is the correct answer. Please note you also need to restart one of those (or both) systemd-udev-settle and systemd-udev-trigger. Restarting just systemd-udevd like Camp said was not enough for me." | 18:25 |
| cehteh | aah udev isnt running | 18:25 |
| GyrosGeier | hdparm -y /dev/sdX ? | 18:25 |
| user___ | You have THREE daemons for ONE task? And they don't know about each other... | 18:25 |
| GyrosGeier | wait | 18:25 |
| GyrosGeier | -z | 18:25 |
| GyrosGeier | wrong keyboard | 18:25 |
| user___ | cehteh: I think you need eudevd? | 18:25 |
| GyrosGeier | udevd is the daemon | 18:26 |
| cehteh | i started udevd which started as eudevd .. i guess its an alias | 18:26 |
| GyrosGeier | udev-trigger tells the kernel to generate hotplug events for all existing devices | 18:26 |
| GyrosGeier | udev-settle waits until all events are processed | 18:26 |
| cehteh | but i had to likk it because its not forked in the background, whats the option forthat? | 18:26 |
| GyrosGeier | so the latter two are tools | 18:26 |
| user___ | cehteh: --daemon | 18:27 |
| GyrosGeier | but generally, rescanning is bus specific | 18:27 |
| GyrosGeier | the udev stuff only deals with stuff that was found during a scan | 18:27 |
| cehteh | just entering a command on that console is pain | 18:27 |
| user___ | yes but he is scanning manually now GyrosGeier | 18:28 |
| user___ | cehteh: if you type blindly, does it lose characters? | 18:28 |
| cehteh | it looses characters or repeats them multiple times | 18:29 |
| user___ | wow | 18:29 |
| cehteh | or enters random conrol characters which are not visible and mess the command up | 18:29 |
| GyrosGeier | that can happen when two shells are attached | 18:30 |
| GyrosGeier | and both try to consume characters | 18:31 |
| user___ | GyrosGeier: or when the web to kvm interface has throttling and one of the parts has resend enabled | 18:33 |
| user___ | which is crap | 18:33 |
| cehteh | ok udevd was running .. i rebooted now to get again a clean state | 18:33 |
| user___ | You can't win this, the hw kvm is crap | 18:34 |
| user___ | cehteh: and all your startup settings are lost? Why do you keep rebooting? | 18:34 |
| cehteh | what startup settings? | 18:34 |
| user___ | This is not windows, it does not fix/break itself | 18:34 |
| user___ | If it did not boot 1st time it won't the following 99 times either. | 18:35 |
| cehteh | i wanted the kernel in a clean state, i tried a lot commands | 18:35 |
| cehteh | rediscover partitions, reload modules | 18:35 |
| cehteh | so replied to the ticket, they asked me whats wrong :D | 18:43 |
| cehteh | ok clean state now, udev is running, | 18:45 |
| cehteh | i somehow suspect udev is at fault | 18:46 |
| GyrosGeier | wait | 18:47 |
| GyrosGeier | your problem is mdadm --scan ? | 18:47 |
| GyrosGeier | that is independent from udev | 18:47 |
| cehteh | mhm | 18:47 |
| GyrosGeier | but it needs to happen after "udevadm settle" | 18:47 |
| GyrosGeier | mdadm scan will generate more udev events | 18:48 |
| GyrosGeier | as the MD devices come online | 18:48 |
| GyrosGeier | so if following code depends on them, do another udevadm settle | 18:49 |
| cehteh | ok took me 1 minute to enter 'udevadm settle' and 'mdadm --assemble --scan' | 18:51 |
| cehteh | no change | 18:51 |
| GyrosGeier | mdadm will not assemble degraded arrays in --scan mode | 18:53 |
| cehteh | the arrays are not degraded | 18:53 |
| GyrosGeier | well, if one of the devices can't be found, they are | 18:53 |
| cehteh | the devices are all there | 18:53 |
| cehteh | only md does not start up | 18:53 |
| cehteh | and no entry in dmesg that it even tries | 18:54 |
| cehteh | da können wir leider nichts machen, das Layout wird aufgrund Ihrer lokalen Einstellungen angepasst. | 18:54 |
| cehteh | Was genau wollen Sie denn mit der KVM machen, gibt es etwas wobei wir behilflich sein können? | 18:54 |
| cehteh | ^ reply :D | 18:54 |
| golinux | cehteh: You might get bounced if you paste too many lines | 18:55 |
| GyrosGeier | needs modprobe md? | 18:55 |
| GyrosGeier | and possible some of the personalities | 18:56 |
| cehteh | GyrosGeier: the modules are loaded | 18:56 |
| GyrosGeier | mdadm --assemble --scan --verbose | 18:58 |
| GyrosGeier | that might also help | 18:58 |
| cehteh | aah | 18:58 |
| cehteh | eww .. it does not scan paritions | 18:59 |
| GyrosGeier | /etc/mdadm/mdadm.conf | 18:59 |
| cehteh | is correct | 18:59 |
| GyrosGeier | there is a DEVICE line there | 18:59 |
| GyrosGeier | if it's commented out or contains "partitions", then it should scan partitions | 19:00 |
| cehteh | ah | 19:01 |
| cehteh | that might be te problem (worekd before do) | 19:01 |
| cehteh | ok rebooting into rescue and fixing that | 19:02 |
| cehteh | GyrosGeier: <3 ... i owe you a beer! that was it. and thanks to everone else helped! | 19:09 |
| GyrosGeier | \o/ | 19:09 |
| cehteh | makes me wonder why it worked before ... | 19:10 |
| GyrosGeier | the initrd also scans for and sets up md devices | 19:10 |
| GyrosGeier | in case root is one | 19:10 |
| GyrosGeier | so this is run twice | 19:10 |
| GyrosGeier | and each has its own copy of the config file | 19:11 |
| cehteh | yeah i usually known that it pulls up any array thats non degraded and finds, suddenly that seems to be not the case anymore | 19:11 |
| GyrosGeier | so I suspect something changed with the initramfs generation | 19:11 |
| cehteh | yes i pretty much narrowed it down to a initrd problem | 19:12 |
| cehteh | Harzilein: ping | 19:12 |
| cehteh | and surprisingly --- all containers came up :D ... just running without a hitch now | 19:14 |
| user___ | And what fixed it? | 19:23 |
| user___ | ok read backlog. DEVICES= eh | 19:26 |
| user___ | re: ticket: cehteh: insist on clarifying it, tell them you can't type certain keys and about the 2 second delay | 19:26 |
| rwp | cehteh, For reference what was the root cause of the problem? The "DEVICE partitions" was something different? | 19:27 |
| cehteh | i had all DEVICES where raid partitons are on listed, but not as wildcard, just the main device | 19:27 |
| cehteh | that woreked before obliviously | 19:27 |
| cehteh | now i just changed it to DEVICES partitions | 19:28 |
| cehteh | was a goodwill attempt to constrain the config .. and still a fault | 19:28 |
| rwp | Good deal that you now have it working! I'll file that tidbit in the back of my head in case it happens again. | 19:34 |
| rwp | Good job to GyrosGeier for that root cause possibility! | 19:34 |
| cehteh | its almost always such an simple fuckup .. still hard to find unless you have an idea what may cause it | 19:38 |
| cehteh | esp no indication of any error until i did the mdadm --assemble --scan --verbose which shown the devices it scanned and failed | 19:39 |
| rwp | There are an infinite number of ways for things to fail. But usually only one way for them to work. | 19:41 |
| rwp | The smallest of details can lead to the biggest problems! | 19:43 |
| Harzilein | cehteh: pong (trying to find your message window on this here emergency irc client ;) | 20:18 |
| cehteh | Harzilein: server works again | 20:20 |
| cehteh | no1 in 1023 days, 23:34:29 | at Sun Apr 28 17:57:00 2024 | 20:23 |
| cehteh | uprecords :D | 20:23 |
| cehteh | that'll be some time | 20:23 |
| Harzilein | nice | 20:30 |
| Harzilein | GyrosGeier++ | 20:30 |
| Harzilein | hi btw :) | 20:30 |
| xrogaan | All the app icons in my application menu vanished | 20:35 |
| xrogaan | they're still present if I use the editor though | 20:35 |
| xrogaan | that's weird. | 20:35 |
| xrogaan | and wiskers show them too | 20:35 |
| user___ | xrogaan: changed permissions recently? bulk changed? | 20:46 |
| xrogaan | nah | 20:53 |
| xrogaan | they're showing fine in other plugins (whisker) | 20:53 |
| xrogaan | it's just not showing in the regular application menu | 20:54 |
| xrogaan | so, probably not the icons themselves | 20:54 |
| user___ | windows remedy: restart the desktop | 21:04 |
| msiism | Is there a way to display the current font and size used on the TTY? | 23:22 |
| fsmithred_ | dpkg-reconfigure console-setup to change it | 23:23 |
| fsmithred_ | oh, look in /etc/default/ | 23:23 |
| msiism | Okay, thanks, will do. | 23:24 |
| * msiism found it. | 23:24 | |
| fsmithred_ | you can just edit that file. Probably quicker. | 23:25 |
| msiism | Yeah, I don't want to change the font. I really just wanted to know what I use. | 23:26 |
| Centurion_Dan | Hi | 23:27 |
| Centurion_Dan | I came across an old installation of Devuan Jessie today on a laptop I setup probably 5 years ago.... Do you think I can still upgrade up to Beowulf and Chimera without it breaking?? | 23:28 |
| gnarface | if you upgrade through ascii first it's theoretically possible | 23:31 |
| gnarface | it has worked on some of my headless machines anyway | 23:32 |
| gnarface | well, up to beowulf anyway | 23:32 |
| msiism | I'd collect all the delicate items from all the release notes before doing that. | 23:32 |
| fluffywolf | Centurion_Dan: it should be fine to beowulf, but are you really sure you want chimaera? | 23:36 |
| fluffywolf | actually, never mind, chimaera should be reasonably usable now. | 23:36 |
| fsmithred_ | check eudev version. You may have to upgrade to a lower version number. | 23:39 |
| fsmithred_ | that's the main gotcha I can think of | 23:39 |
| fsmithred_ | is eudev in jessie? | 23:40 |
| fsmithred_ | Centurion_Dan, you might do well to skip past ascii | 23:42 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!