systemdlete | does hibernate work, and if so, is it reliable (vs buggy)? | 00:57 |
---|---|---|
debdog | systemdlete: what devuan release? | 00:57 |
systemdlete | daedalus | 00:57 |
systemdlete | (hi debdog!) | 00:58 |
debdog | o/ | 00:58 |
systemdlete | I'm about to try it on a PC with daedalus boot | 00:58 |
systemdlete | but just wondering if there are major known issues I need to know | 00:58 |
debdog | since daedalus is pretty new, you prolly doing field work | 00:59 |
systemdlete | ok... | 00:59 |
systemdlete | good to know at least that much | 00:59 |
debdog | but I've heard the situation has improved over chimaera | 01:00 |
systemdlete | looks like I need to add some software to make that option appear in xfce status UI | 01:00 |
* debdog has closed his laptop's lid once on daedalus and the system came back after opening it again | 01:01 | |
systemdlete | are you sure that isn't suspend? | 01:01 |
debdog | hehe, what's the difference | 01:01 |
systemdlete | as in suspend to ram as opposed to suspend to disk? | 01:01 |
debdog | it's realted | 01:01 |
systemdlete | suspend to ram works here | 01:01 |
debdog | IIRC GPU drivers are the ones causing the most troubles | 01:02 |
systemdlete | My daedalus install does not have a hibernate option in the shut down options | 01:04 |
systemdlete | the debian wiki suggests installing a gnome button exxtension | 01:04 |
systemdlete | idc, it's a test box. | 01:05 |
debdog | hehe, that's the spirit | 01:06 |
debdog | being like a child | 01:06 |
systemdlete | yes. It's amazing how our formalized educational system has managed to kill that innocent instinct to be curious. Everything is "don't touch unless you get permission first" | 01:07 |
systemdlete | but that's getting OT | 01:07 |
systemdlete | and I don't want to face an arrest here | 01:07 |
gnarface | hibernate should work on everything, in theory. it doesn't require special hardware features like suspend does, it just requires a swap partition equal to or larger than the size of your physical ram | 01:14 |
gnarface | as long as you have the pm-utils package, you should be able to test it by calling /usr/sbin/pm-hibernate directly (as root) | 01:14 |
systemdlete | the hibernate option in the options menu is grayed out. xfce4 wiki indicats this is because os does not support it | 01:15 |
systemdlete | ah... pm-hibernate | 01:16 |
gnarface | on some older PCs, it's proper functionality may be contingent upon a particular bios setting to do with (memory foggy here) something like "memory hole remapping at 15k" being set correctly (on or off i don't even remember anymore, but it was important) | 01:16 |
gnarface | *its | 01:16 |
gnarface | the xfce4 graphical option is probably just greyed out because you're missing a particular xfce package and/or setting | 01:17 |
gnarface | note though that on lots of hardware this won't actually save you any time booting up | 01:18 |
* systemdlete is install pm-utils right now | 01:18 | |
gnarface | (especially if you're doing something like restoring a 16GB ram image from spinning rust) | 01:19 |
systemdlete | is FX8350 "spinning rust"? | 01:19 |
gnarface | no i mean harddrives | 01:19 |
systemdlete | anyway, I'm not trying to save time. | 01:19 |
gnarface | just be aware that it's not always obvious when it's actually working, because it typically causes a LLOOOONGG black-screen pause during boot up | 01:20 |
systemdlete | I just want to be able to suspend my system to disk, boot a thumbdrive and do something with it, then resume whatever I was doing beore | 01:20 |
systemdlete | before | 01:20 |
gnarface | resuming from a thumb drive is going to be even slower | 01:20 |
systemdlete | gnarface, yeah... when the boot is trying to find a resume image | 01:20 |
systemdlete | NO! | 01:21 |
systemdlete | resuming from the disk. | 01:21 |
systemdlete | What I am saying is... | 01:21 |
gnarface | it's gotta literally physically load the entire swap partition back into ram, not just "find it" | 01:21 |
systemdlete | I want to shudown my hard-disk based system to disk, boot up something else entirely, maybe off a USB, do stuff with that (but not touching the suspended disks), then quit the USB stuff, and then RESUME the hard-disk based system | 01:22 |
gnarface | so whatever the raw speed of reading 16GB of data off your disk is, in whatever the default speed the bios syncs the drive up as, that's the new minimum boot time with resume from hibernate | 01:22 |
systemdlete | I think these are 5400 drives, one might be 7200, irdr now | 01:22 |
systemdlete | and to make things more awful, this is 32G | 01:22 |
gnarface | oh wow. that's gonna be... educational. i'm just trying to warn you against false positives; it may seem like it's hung when it's actually working. just be patient. | 01:23 |
systemdlete | I'm not planning on doing this a lot... although, depending how things turn out I might change my mind | 01:23 |
fsmithred | why not just do whatever on the usb in a chroot? | 01:23 |
fsmithred | then you don't have to reboot | 01:23 |
systemdlete | gnarface, you are asking me to be patient. :D | 01:23 |
systemdlete | fsmithred, this usb thingy seems to require taking over the entire hardware | 01:23 |
gnarface | heh, fair enough | 01:24 |
systemdlete | otherwise, yeah, I'd be doing it in a VM anyway | 01:24 |
systemdlete | I am pretty sure I've seen suspend-to-disk/resume-from-disk work in at least chimaera | 01:24 |
fsmithred | hibernate to disk or suspend to ram | 01:25 |
fsmithred | and then some kind of hybrid that I don't know | 01:25 |
systemdlete | right. But even after installing pm-utils, the hibernate option is still disabled for some reason. | 01:26 |
fsmithred | I get variable results with suspend to ram in chimaera and daedalus. | 01:26 |
systemdlete | I'm getting good results here with either. | 01:26 |
fsmithred | you have a swap partition as big as your ram? | 01:26 |
systemdlete | I'd better! | 01:27 |
systemdlete | but I'll double-check | 01:27 |
fsmithred | well, not having that would be a reason for hibernate to be greyed-out | 01:28 |
systemdlete | Usually, when I create a system, I make sure my swap is at least 2x my RAM | 01:28 |
systemdlete | oh? | 01:28 |
systemdlete | the xfce wiki only says that the greyed-out button is if the OS does not support it | 01:28 |
fsmithred | what does that mean? | 01:29 |
systemdlete | I think that's referring to the absence of supporting software | 01:29 |
systemdlete | pm-utils maybe | 01:29 |
systemdlete | but anyway, fsmithred, I think you nailed it. | 01:30 |
systemdlete | When I built the box originally, it only has 16G. But I've replace it with 32G. So the original 20G of swap is not enough. I will add some... | 01:30 |
gnarface | systemdlete: you have acpi installed too, right? | 01:31 |
systemdlete | can the swap be spread over multiple devices? | 01:31 |
systemdlete | gnarface, hold on | 01:31 |
gnarface | swap can, but i have no idea if that will sabotage hibernate or not | 01:31 |
gnarface | never tested it, dunno if it has to be contiguous | 01:32 |
systemdlete | (that's wwhat I meant...) | 01:32 |
systemdlete | OK, just installed acpid. No change with the options. But I see I am short of swap. | 01:35 |
systemdlete | So I am looking for some more space... | 01:36 |
systemdlete | I feel like somebody calling the Georgia Secretary of State and telling them I just need 12G of additional swap so I can win | 01:36 |
systemdlete | I've got gobs of hardware here, including disks. No reason for this other than I am chinsy. | 01:37 |
gnarface | hehe, well there's always USB... | 01:37 |
systemdlete | A test box deserves some large disks, doesn't it? | 01:37 |
gnarface | sure | 01:37 |
systemdlete | Of course, that will eat the rest of the day... | 01:37 |
gnarface | i don't actually have any reason to believe resuming from suspend on a usb swap partition will actually work | 01:37 |
gnarface | it might depend on the motherboard | 01:38 |
fsmithred | oh, that might be my resume problem on this laptop! I'm running off a usb. | 01:38 |
fsmithred | thanks, gnarface | 01:38 |
systemdlete | eh. I'll stick with real disk | 01:39 |
gnarface | fsmithred: only thing i can think of that might fix it, just maybe, is if you go into the bios and find a setting for something like "usb legacy mode" enable it | 01:39 |
fsmithred | maybe. I'm actually ok with not suspending this laptop. | 01:40 |
gnarface | usb legacy mode does something analogous for keyboards, i don't know specifically if it'd work for block devices too but it'd be worth a try | 01:40 |
fsmithred | I've got a minimal lxqt install on an old EEE netbook, and suspend works corretly on that. I didn't have to mess with settings at all. | 01:41 |
fsmithred | I think I have a hardware problem, because I sometimes get syslog messages asking if the usb cable is bad. | 01:42 |
fsmithred | And that coincides with getting I/O errors in a terminal | 01:43 |
systemdlete | ok, found some space. And that's without help from Rathensburger | 01:43 |
systemdlete | (just to continue the OT silliness) | 01:43 |
systemdlete | okay... so the hibernate option is no longer grayed out. Reason: I had to set the correct UUID in /etc/initramfs-tools/conf.d/resume *and* add "resume=UUID=same_correct_uuid" to /etc/default/grub default boot line | 02:14 |
systemdlete | then run update-grub, update-initramfs and reboot. | 02:14 |
systemdlete | now, for the test... (drumroll) | 02:14 |
DelTomix | we are all counting on you | 02:15 |
gnarface | good luck! | 02:15 |
systemdlete | holy! | 02:16 |
systemdlete | it worked! | 02:16 |
systemdlete | and it really wasn't too long | 02:16 |
systemdlete | maybe 10-15 seconds or so down and up again | 02:16 |
systemdlete | (aside from the hw boot itself I mean) | 02:17 |
gnarface | nice! | 02:18 |
systemdlete | so to answer my own question: the resume= parameters in those 2 files must point to the same UUID, and I'm guessing it must be just that one. | 02:18 |
gnarface | your bios must be smart enough to not bring the disks up in PIO mode :-p | 02:18 |
gnarface | it's not always a given | 02:18 |
systemdlete | and maybe that's a bios option? | 02:18 |
gnarface | i've never seen it off by default, but i've seen some raid cards that couldn't set dma speeds right until the linux kernel loaded | 02:19 |
systemdlete | yeah. Now that I'm thinking about it... I do seem to recall something about hibernation and sleep not always working right because of that | 02:19 |
gnarface | loading 32GB in PIO mode would take so long it would certainly appear frozen | 02:20 |
systemdlete | well, I did get the blank screen of who-knows-how-long, but like I said, it only took about 10 seconds or so | 02:20 |
gnarface | that's good. those are good resume times. | 02:20 |
systemdlete | and for as often as I am currently planning to do this, I think it will be OK. | 02:21 |
systemdlete | actually, maybe I SHOULD hibernate the box since it's a test box that I don't use all the time. | 02:21 |
gnarface | now the real test; make sure none of the drivers went insane while hibernating. (nvidia's linux drivers are especially bad at waking up from various states) | 02:21 |
DelTomix | now you have me thinking I should set this up on beaker (my test box) | 02:22 |
gnarface | hmm, i wonder if he tripped on a driver fault | 02:33 |
gnarface | systemdlete: driver issue? | 02:49 |
systemdlete | gnarface, driver issue... ? | 02:52 |
gnarface | systemdlete: oh, your irc connection ping timed out shortly after i was warning you about nvidia driver instability on resume so i thought maybe you got hit by it headlong | 02:54 |
gnarface | *head-on | 02:55 |
systemdlete | oh | 02:55 |
systemdlete | yeah, I've been having some odd problem for a few days now where my modem quits the DHCP lease (maybe at the ISP, idk) | 02:55 |
systemdlete | I reboot my openwrt router and that fixes it. | 02:56 |
systemdlete | annoying | 02:56 |
systemdlete | last thing I wrote was I don't see any options about PIO mode. Then it quit | 02:57 |
systemdlete | so anything after that I didn't get | 02:57 |
systemdlete | gnarface, is PIO only an option for SATA perhaps? | 03:17 |
systemdlete | idk much about this. | 03:17 |
gnarface | systemdlete: no, PIO is also a PATA thing. you might see it in the BIOS under the disk setup parts that you don't have to mess with anymore on modern computers, but sometimes would have to go into that section to set up cylander counts and stuff on older machines back in the 90's | 03:19 |
systemdlete | oh, like LBA option, which is nowadays pretty much standard | 03:19 |
systemdlete | (I remember those!) | 03:19 |
gnarface | yea. auto-detect always works for those these days. but way back in the day you'd have to sometimes actually go set stuff manually in that section. you will also see supported PIO modes of a given disk listed in the output of hdparm -I | 03:20 |
systemdlete | anyway, is there any advantage to switching from IDE to AHCI mode? | 03:20 |
gnarface | speed, theoretically, but it can cause compatibility issues in some cases | 03:20 |
systemdlete | and can it be done safely, without damaging my data? | 03:20 |
gnarface | probably, unless it's a RAID array | 03:21 |
systemdlete | well, actually there are 2 disks in all of my machines as RAID 1 members | 03:21 |
systemdlete | but I run software RAID 1 (mdadm) | 03:21 |
systemdlete | I don't use the bios for RAID | 03:21 |
gnarface | i couldn't say if it will affect software raid or not | 03:21 |
gnarface | but in bios raid setups i've usually seen AHCI set in opposition to RAID mode while IDE mode is sometimes listed as equivalent | 03:22 |
gnarface | i wouldn't expect it to corrupt any data, i'd expect it to just not boot, but like i said i'm not certain | 03:23 |
gnarface | and to be clear, don't worry about that PIO thing. it's basically just the legacy fallback mode you get if DMA/UDMA fails. | 03:24 |
systemdlete | this bios lists all 3 as options for each individual dirve | 03:24 |
gnarface | if you loaded a 32GB hibernation image in 10s you're definitely using (U)DMA | 03:24 |
systemdlete | I think it is awesome. | 03:24 |
gnarface | hmm, does the motherboard manual mention any definitions? usually the limitations are bios specific | 03:25 |
systemdlete | eh... I think I might actually have the manual here. | 03:25 |
gnarface | hmm... | 03:25 |
systemdlete | It's a slightly different board than the one that came with this manual | 03:26 |
systemdlete | but I think they are probably the same bios | 03:26 |
gnarface | ok, feel free to use AHCI for software raid, according to the same question answered here on tomshardware: https://forums.tomshardware.com/threads/ahci-or-raid-in-bios-if-adding-software-raid-later.3037512/ | 03:27 |
gnarface | that matches what i'd guess, so i'd say try it | 03:27 |
systemdlete | ah, good old tom's hardware | 03:27 |
gnarface | since you're not booting from the raid array, just from a single disk, the bios won't care | 03:27 |
systemdlete | what I can do is disable one drive and and see if the thing still boots. Then I can sync the other drive. | 03:28 |
systemdlete | if not, then I can boot off the other and sync the other | 03:28 |
systemdlete | set to [IDE] when you want to use the serial ATA hard disk drive as Parallel ATA physcial storage devices | 03:35 |
systemdlete | set to [RAID] when you want to create a RAID configuration from the SATA hard disk drives | 03:35 |
systemdlete | set to [AHCI] when you wnt th SATA hard disk drives to use the AHCI (Advanced HOst Controller Interface). | 03:36 |
systemdlete | (sorry for the bad typing. you get the idea) | 03:36 |
gnarface | hmm, i have to assume that they mean for hardware raid in that case though | 03:36 |
systemdlete | yes, I think so. | 03:36 |
systemdlete | I tried a raid card once a few years ago. It was such a nuisance to set up that I opted for mdadm instead. (mdadm had just replace the older raid tools then) | 03:37 |
systemdlete | I think maybe c. 2007-2010 or so? | 03:38 |
systemdlete | in fact, iirc, it required the use of a windows tool (came with it). | 03:38 |
systemdlete | wow. Not only was it able to boot under ahci, but it did so from resume! | 03:43 |
systemdlete | now, if I hot-plug the other disk, will it sync? | 03:44 |
gnarface | uh... i wouldn't gamble on SATA hotplug being actually implemented and available for internal SATA connections | 03:45 |
systemdlete | actually, it works | 03:46 |
systemdlete | I have a 4 bay unit that supports hotplug | 03:46 |
systemdlete | I don't do it often, but when I have, it has always worked | 03:46 |
gnarface | ah, i've heard mixed opinions about whether that is supposed to be a requirement of the spec, but i've almost never seen it actually implemented for motherboards' internal ports, only the "esata" back panel connectors | 03:47 |
systemdlete | I think I've heard the same muddy message as well. | 03:51 |
systemdlete | I'm pretty sure these 4-bay units were designed for use with hotswapping/hotplugging. | 03:52 |
systemdlete | but I think these units have no real smarts themselves. | 03:52 |
systemdlete | I have 2 kingwin 4-bay units, one on my test box and the other on a different box. | 03:53 |
systemdlete | they have SATA power and data connectors in the back of each slot, and those just get connected to the mainboard. There's no contoller card to plug in. And I don't see any internal card on either. | 03:55 |
gnarface | i honestly couldn't recommend trusting them to be hotplug safe without getting a more authoritative answer about it. i've heard people even succeeding at hotplug removal of pata disks if they're lucky and fast... once or twice before it fries something | 03:56 |
systemdlete | these are sata not pata disks | 03:57 |
gnarface | yea i know, but my point still stands, as i'm not sure you can trust sata hotplug to always be actually a furnished feature everywhere | 03:57 |
systemdlete | and I have not tried hotswap or hotplug with IDE drives when I had them. I was aware that they could be problematic | 03:58 |
rwp | Always use AHCI if available. It's the right answer. | 04:03 |
systemdlete | in the bios, under each sata drive slot, there are options, including things like PIO (0-4 and auto). I have auto set for nearly everything in the bios. | 04:03 |
rwp | IDE devices were generally not hotswappable. | 04:04 |
rwp | All SATA/SAS device hardware is hotswappable by specification. And I have often hotswapped them. | 04:04 |
systemdlete | https://web.archive.org/web/20140908152103/http://wdc.custhelp.com/app/answers/detail/a_id/941/~/hot-swap-or-hot-plug-wd-sata-drives | 04:10 |
systemdlete | (rwp, yes, that's what I just read at the link) | 04:11 |
systemdlete | (and my drives are mostly WD) | 04:11 |
systemdlete | switching modes from IDE to AHCI did not do any damage that I can tell. | 04:11 |
systemdlete | I think the pins on the connectors are designed so that they are either all in or not at once. The only caveat is they warn not to use molex power connectors. | 04:12 |
systemdlete | rwp: I have IDE hotswap bays that were able to do it. | 04:13 |
systemdlete | but that's because they were designed for that (and I paid a good bit of money for them, iirc) | 04:13 |
systemdlete | this hibernate and suspend thing is so wicked! I love it. | 04:22 |
rwp | systemdlete, IDE was not *generally* hotswap. But yes there were setups that would do it. But all SATA/SAS hardware is hotswappable. | 08:02 |
rwp | Welcome to the joys of suspend and hibernate! | 08:03 |
systemdlete | rwp: ty! | 08:04 |
systemdlete | I can't get over how well it works. It works even if I am running VMs! | 08:04 |
systemdlete | restores all of it, no crashes | 08:04 |
strato | I got this problem: | 10:20 |
strato | 2 Nearly identical Beowulf fully upgraded, (with X), start to reboot | 10:20 |
strato | randomly. I tryed chimera and daedelus on one of them, same problem. | 10:20 |
strato | Then I installed devuan_chimaera_4.0.0_i386_server.iso (without X) | 10:20 |
strato | on one of them and it has been stable for 3 days. | 10:20 |
strato | The other still reboots randomly within few hours. | 10:20 |
strato | Any sugestion ? | 10:20 |
jaromil | It is very likely an hardware problem. I had such problems in the past and changing RAM or even mo/bo solved | 10:25 |
strato | But 2 servers nearly at the same time? | 10:25 |
onefang | Power glitches maybe? | 10:46 |
strato | I thought so, but now they are running two different location. | 10:49 |
strato | I have a suspision, Could it be conectet to a upgrade, but then agian, I got other servers with nearly the same config. | 10:50 |
gnarface | strato: the first thing i would suspect is the power supply or the power source | 12:31 |
gnarface | weak and failing power supplies do that, and so do computers on flaky power grids | 12:33 |
gnarface | could be the video cards failure or a regression in the video drivers but the behavior you describe is more like a power issue and disabling X would also make the video card use less power (by a lot, in some cases) | 12:34 |
gnarface | a few years ago i started having a problem with my computers randomly rebooting too. i gave them all on battery backup power and that stopped. | 12:36 |
gnarface | *put them all on... | 12:37 |
strato | Interresting: I will demount X if this continue, and see if that's the issue. | 12:44 |
gnarface | strato: depending on the generation of hardware, overheating could also be the issue. a thorough check of the fans is probably in order too. | 15:30 |
buZz | strato: 32bit? :O | 15:31 |
buZz | just how old -are- these servers :D | 15:31 |
buZz | but also , i highly recommend not running any GUI on servers that are actually servers | 15:31 |
buZz | also; if its a decent server, you might be able to see a event log in IPMI/whichever about the reboot reasons | 15:32 |
Xenguy | Yeah, if it's a 'server' per se, X should not be running | 15:37 |
gnarface | yea, but it also really shouldn't be causing it to reboot. i've seen lots of video driver and video hardware failures, and they sometimes lockup the system, but they usually just cause rendering anomalies, rarely reboots. | 15:39 |
gnarface | reboots are usually heat or power | 15:39 |
gnarface | too much or too little, respectively | 15:40 |
buZz | i've had some old poweredge that kept rebooting because diskcontrollers battery was empty | 15:40 |
buZz | yeeted the whole battery, no backup anymore, worked fine since | 15:40 |
gnarface | ... and happening across 3 releases strongly suggests it's not the drivers | 15:41 |
gnarface | happening to two of the same models of the same age, now that could be hardware failure | 15:41 |
gnarface | but if this was in california i'd definitely bet on it just being the shitty power grid doing this, and suggest you waste no time getting UPSes | 15:42 |
gnarface | the cheap little UPSes i get have a digital counter on them and they've been counting up at a rate of dozens of events per year for the past several | 15:44 |
gnarface | but none of my computers spontaneously reboot anymore | 15:44 |
gnarface | not even the ones with nvidia cards | 15:44 |
buZz | yeah many PSUs dont really like when your 240V suddenly drops to 200V | 15:55 |
buZz | or equivalent for edison nations | 15:55 |
bb|hcb | buZz: California is on 110V ;) BTW. Most PSUs are designed to work 100-240V, so 200V instead of 220V is a lesser problem. Power suges and/or drops are a bigger problem | 16:31 |
buZz | bb|hcb: yeah most nations that fell for Edisons stolen tech went for 110V | 16:32 |
bb|hcb | ... and the instability, especially in oldish hw is due to drying capacitors | 16:36 |
buZz | could be, sure | 16:37 |
al1r4d | hello | 16:38 |
gnarface | hello al1r4d, you know you don't need permission to ask questions right? | 16:47 |
al1r4d | gnarface, nope, i dont want to ask, just say hello | 16:47 |
al1r4d | I'm addicted to openbsd haha | 16:48 |
golinux | Which is offtopic here | 17:05 |
user24037 | hi; how should i use custom dns resolver? setting in my router is not an option | 19:19 |
gnarface | user24037: you should be able to set it in /etc/resolv.conf | 19:20 |
gnarface | well, it could get ignored or overwritten by networkmanager or the like | 19:21 |
gnarface | if you're using a graphical network setup tool it should have that feature | 19:21 |
gnarface | you might have to switch to manual configuration from dhcp | 19:21 |
user24037 | OK. i had been looking at generating a resolv.conf with the resolvconf service but it wasn't working. maybe manual configuration makes more sense. | 19:23 |
gnarface | user24037: the resolvconf package just lets you set the dns in /etc/network/interfaces instead, but that would also conflict with things like networkmanager | 19:23 |
user24037 | i don't think i have networkmanager. i used to run wicd but that got depreciated | 19:24 |
gnarface | whatever you used for dhcp should have created a working /etc/resolv.conf then | 19:25 |
gnarface | you should be able to just edit it, as long as dhcp is disabled | 19:25 |
fsmithred | If you do have network-manager, it will be polite enough to leave you a note when it clobbers your custom settings in /etc/resolv.conf | 19:30 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!