EHeM | As debdog noticed, *silently* is the key word; many channels won't let you say anything without some flavor of verification, but most will at least hint that you're being ignored. | 00:14 |
---|---|---|
MinceR | that's the ircd's fault | 00:14 |
EHeM | Anyway, sometime in the past 2 weeks the repositories got slightly broken, the stable entries no longer work (fine, many people dislike them, but for some they're a very handy feature, like all features). | 00:15 |
fsmithred | slightly broken, then a little different broken, then fixed | 00:22 |
EHeM | fsmithred: "Err:9 http://deb.devuan.org//merged stable Release", " 404 Not Found [IP: 131.188.12.211 80]"; that is broken and still broken (just retried now). | 00:28 |
fsmithred | worked yesterday. Maybe it's broken again? | 00:29 |
fsmithred | I'll try | 00:29 |
golinux | Try another mirror | 00:29 |
EHeM | I've tried several mirrors over the past few days and it has remained broken. | 00:29 |
golinux | http://pkgmaster.devuan.org/mirror_list.txt | 00:30 |
fsmithred | deb.devuan.org works here | 00:32 |
fsmithred | I don't know what mirror I'm hitting | 00:32 |
EHeM | Well, ftp.rrze.uni-erlangen.de, megumin.packet-gain.de, devuan.bio.lmu.de and 5.196.38.18 all fail the test. | 00:32 |
EHeM | "deb http://us.deb.devuan.org/merged/ stable main" is the line from sources.list. | 00:33 |
golinux | I just updated here and np | 00:34 |
fsmithred | pretty sure the us prefix does nothing yet | 00:34 |
golinux | Yeah, we don't have a us mirror yet. | 00:34 |
EHeM | fsmithred: Given how I'm hitting servers in Germany, I think you're right, but I do /try/ to reduce my use of bandwidth. | 00:34 |
golinux | IIUC, deb.devuan.org will default to one close to you | 00:35 |
EHeM | This time got mirror.checkdomain.de, fail. | 00:36 |
golinux | Let me check something | 00:37 |
golinux | Evilham's mirror status script says everything is OK. | 00:38 |
golinux | I think I found it | 00:38 |
EHeM | Further error message says missing Release file. | 00:38 |
golinux | http://deb.devuan.org//merged . . . it's those // | 00:39 |
golinux | Try with only one | 00:39 |
EHeM | Nope. | 00:40 |
fsmithred | I see a release file at 131.188.12.211 | 00:41 |
fsmithred | 131.188.12.211/devuan/devuan/dists/2.0/ | 00:41 |
EHeM | Added "131.188.12.211 us.deb.devuan.org pkgmaster.devuan.org" to /etc/hosts and fail. | 00:43 |
rrq | note that "deb.devuan.org" is merely a multi-host dns, and there's no resolution cleverness other than settling for a single one at "random" at each request. | 00:46 |
rrq | also if it's a "jessie" update: jessie has got archived, so jessie-updates and jessie-backports are not there anymore. | 00:48 |
rrq | or are they? some mirrors have them; maybe it's too early int the morning for me... | 00:51 |
fsmithred | jessie-backports is gone, I know that much | 00:52 |
fsmithred | I believe jessie updates is still working. I didn't comment it out in my sources.list. | 00:54 |
golinux | Centurion-Dan said updates are gone - I got bit by that. | 00:54 |
fsmithred | Hit http://security.debian.org jessie/updates InRelease | 00:55 |
fsmithred | nope, that's jessie-security | 00:56 |
golinux | jessie-updates aren't | 00:57 |
fsmithred | oh, I did comment out jessie-updates | 00:58 |
EHeM | One reason I use "stable" is that eliminates the need to update sources.list with each update; if APT had some sort of variable substitution I might consider making use of that. | 01:02 |
golinux | Problem is that Debian stable and Devuan stable are not always in sync | 01:03 |
fsmithred | EHeM, that wou... | 01:03 |
fsmithred | yeah, that | 01:03 |
golinux | So when buster goes stable and beowulf isn't . . . oops | 01:04 |
Evilham | from my mirror checker: | 01:06 |
Evilham | NOTE: This checker only compares propagation and up-time across the mirror network. Not correctness of reference server (pkgmaster.devuan.org). | 01:06 |
EHeM | So anyway, it got br0k3n in the past 2 weeks and it would be nice if it could be made to work again... | 02:01 |
fsmithred | did you try changing stable to ascii? | 02:03 |
EHeM | That does allow APT to verify the data it is getting and produces no error. | 02:03 |
fsmithred | it's also the way that's recommended and supported | 02:04 |
EHeM | "stable" is a standard supportted dist too; sure there is dislike of it, but used with care it is quite handy. | 02:06 |
EHeM | For a large installation it might be handy to have one system on stable, which will alert you when an update is ready. | 02:06 |
fsmithred | I don't understand what benefit you see in it | 02:07 |
fsmithred | what update? | 02:07 |
fsmithred | regular package updates, or update from one release to the next? | 02:07 |
EHeM | Update from one release to the next. | 02:07 |
fsmithred | so you have a time bomb | 02:07 |
fsmithred | you can't take that upgrade when it comes | 02:08 |
EHeM | If you handle it improperly, yes it is a timebomb, if you use care it is quite handy. | 02:08 |
fsmithred | how are you going to keep the devuan packages separate from the debian packages that are upgrading to the next release? | 02:09 |
fsmithred | and what are you going to do with all the version conflicts? | 02:09 |
golinux | He's not listening. Save your breath | 02:09 |
fsmithred | how do you think you're going to handle it properly? | 02:11 |
fsmithred | take your time, I'll be back in five minutes | 02:11 |
EHeM | fsmithred: I'm only using Devuan's repositories (someone who used a repository targetted at Debian would have a trickier issue), version conflicts help ensure everything remains on a single Devuan version. | 02:12 |
EHeM | The concerns about breakage have been around more than a decade, yet no problems have ever cropped up for me. | 02:14 |
fsmithred | but you're using a symlink | 02:15 |
EHeM | Yes. So? | 02:15 |
fsmithred | and when debian buster goes stable, the symlinks will change | 02:16 |
fsmithred | but the devuan packages won't | 02:16 |
EHeM | There could be symbolic links at several places in the repository, right now I'm too small to a local mirror. | 02:17 |
fsmithred | use "ascii" and when buster goes stable, you can upgrade to "beowulf" | 02:18 |
fsmithred | then all your package versions will match | 02:18 |
fsmithred | then you won't be mixing current stable with testing (or when it happens, oldstable with stable) | 02:18 |
EHeM | I'm pretty sure the Packages file points to the pool directory which then contains everything, the symbolic links would be involved in the Packages file which unless you're deliberately creating a mixed Packages file won't cause problems. | 02:19 |
fsmithred | we are merging two different repositories | 02:19 |
fsmithred | ours and debian | 02:19 |
fsmithred | and filter out the debian packages that we've re-packaged | 02:19 |
EHeM | And unless you're deliberately creating a stable repository which isn't a symbolic link to ascii or beowulf, stable will remain one version. | 02:20 |
EHeM | I've got a tested backup strategy, worst case is I'm forced to restore a system (potentially a VM) and will then no longer use stable. | 02:32 |
EHeM | The warnings about using "stable" go back a decade, yet no breakage has every shown up. | 02:33 |
EHeM | ^every^ever | 02:34 |
fsmithred | the warning of using "stable" that I know about only go back a year or two | 02:39 |
fsmithred | in devuan | 02:39 |
fsmithred | and they came from people who got burned by it. Not theory. | 02:40 |
fsmithred | oh yeah, I do recall warnings about the upgrade from sarge to etch. I was one of the ones who got burned by that. | 02:49 |
EHeM | What sort of breakage did you observe? | 02:50 |
fsmithred | xfree86 to xorg was a mess | 02:50 |
fsmithred | I ended up reinstalling | 02:50 |
fsmithred | it was a kid's computer, so no backups and nothing important lost | 02:51 |
EHeM | I think what I ended up doing was sticking with XFree86 for a bit until I was certain I had all the correct packages selected and *then* let X get upgraded. | 02:52 |
fsmithred | what I think you'll see with devuan is version conflicts between core packages | 02:55 |
fsmithred | example, util-linux and libc6 | 02:55 |
fsmithred | one is a devuan package the other not | 02:55 |
fsmithred | and I don't see any "stable" in /merged. That might explain why your apt can't find the Release file, but I can see it's there in my web browser. | 02:58 |
EHeM | If breakage shows up, I'll be forced to admit my strategy was flawed, but I've still yet to have any problems. | 02:59 |
fsmithred | did you do this for jessie to ascii? | 02:59 |
EHeM | As it was working 2 weeks ago I'd like for it to return to a similar state. | 02:59 |
fsmithred | how was it 10-12 months ago? | 03:00 |
EHeM | fsmithred: Use "stable" in /etc/apt/sources.list and do the update from Devuan jessie to Devuan ascii? Yes, worked fine. | 03:00 |
fsmithred | you may have gotten some unsanitized packages that way. | 03:01 |
EHeM | There aren't any traces of such that I can find (you're welcome to suggest extra places to look). | 03:08 |
golinux | EHeM: This might make it clearer: https://dev1galaxy.org/files/amprolla.png | 03:13 |
fsmithred | you'd have to compare your package list to a list of packages that get modified | 03:14 |
fsmithred | you can do whatever you want, but there's no benefit to it and there is some risk | 03:14 |
golinux | Experience it the best teacher . . . | 03:15 |
golinux | it > is | 03:15 |
golinux | You wasted hours to no avail | 03:15 |
golinux | More than some risk. Besides different version numbers, won't there be the possibility of no filtering of packages needing systemd? | 03:19 |
golinux | I'm not sure how the redirect is set up. I imagine it would be to buster not resting. | 03:19 |
golinux | oops - testing | 03:19 |
fsmithred | as I said, I see no "stable" or "testing" symlinks in /merged | 03:24 |
fsmithred | if they were there two weeks ago, then maybe someone removed them because buster is about to go stable | 03:25 |
EHeM | golinux: Experience is the best teacher, and perhaps I'll have a bad experience with stable in the future, but until then I've got more than a decade and no problems. | 03:41 |
EHeM | fsmithred: That brings the question of why those aren't being dealt with via scripts? | 03:42 |
fsmithred | I haven't looked at the code in amprolla, so I don't know what script there are or aren't. | 03:43 |
EHeM | The needed feature is some level of macro substitution in sources.list, add the ability to set current="ascii" and then substitute "ascii" for "${current}" in a bunch of places and it becomes less of a headache. | 03:45 |
EHeM | In the mean time continuing to support a handy symbolic link supportted by all Debian's variants for over a decade is better. | 03:46 |
EHeM | Is there somewhere else I need to report that things got broken? | 03:47 |
fsmithred | probably not. This conversation will be seen. | 03:49 |
golinux | EHeM: Some reading for you. Have at it: https://git.devuan.org/devuan-infrastructure/amprolla3 | 03:50 |
djph | wait, wasn't amprolla just '2' recently? | 03:50 |
fsmithred | nope. 2 was skipped | 03:51 |
fsmithred | an unfinished project | 03:51 |
fsmithred | 3 is different, so it's 3 | 03:51 |
djph | ah | 03:52 |
djph | perhaps I'm recalling the discussion of v2 then | 03:52 |
golinux | Ut was never in production iirc | 03:52 |
fsmithred | you do | 03:53 |
golinux | That was a while ago. | 03:53 |
djph | i have a weird perception of time, I can go 8 months and call it "like last week, right?" | 03:54 |
* rrq gets reminded of a dejavu he wants to have | 04:00 | |
djph | rrq: don't tempt the Matrix overlords. | 04:03 |
fsmithred | uh-oh, did they cut us off? | 04:11 |
fsmithred | what's 22 minus 15? | 04:12 |
djph | 7? | 04:12 |
fsmithred | I was getting bumped off-line about every hour or two | 04:12 |
fsmithred | and it hasn't happened once since I put in the new router 7 hours ago | 04:12 |
EHeM | Funny how working network hardware makes your network work better. | 04:14 |
djph | ^ | 04:14 |
fsmithred | I only noticed it in here. It would go out for a minute or less. | 04:16 |
* gnarface wasn't ignoring anyone, just wasn't actually here to respond | 06:25 | |
gnarface | tell debdog sorry if he comes back | 06:25 |
Evilham | fwiw: amprolla docs says stable should be a symlink to current stable (aka ascii atm) | 07:24 |
Evilham | so, even if (when) debian releases before devuan does: stable means ascii | 07:24 |
Evilham | no version conflicts | 07:24 |
Evilham | and of course ascii / beowulf track release names, not "stable" or "testing" | 07:27 |
Evilham | (aka, they track stretch / buster respectively) | 07:27 |
Evilham | that being said: ineed, stable should be a thing, it was last Sunday, it wasn't on Monday, it was on Wednesday it's not today | 07:27 |
onefang | That's a somewhat unstable stable then. | 07:30 |
Evilham | yup. | 07:30 |
systemdlete | once a week, on Sunday mornings, the system runs mdadm check on raid arrays. I happened to be around when all the lights on my disk array came on, so I wondered what was up. I figured out that it was the weekly check on raid, but I noticed that 2 of 8 raid devices were not getting checked. | 12:45 |
systemdlete | I requested check on both of them, and the system did what it should. But I wonder how does mdadm or the raid subsystem decide which arrays to check? | 12:46 |
systemdlete | I searched for differences in the /sys/block/md*/md files, and the only distinguishing file was the "last_sync_action" -- the two devices which were not checked showed "none", but after I requested the check they too showed "check" | 12:47 |
systemdlete | It's a bit odd. I looked at the /etc/mdadm config files. I did not notice anything that called out the difference. | 12:48 |
systemdlete | Please note that it is "sync_action" not "last_sync_action" that responds to the requests to run check or repair | 12:51 |
systemdlete | I am only slightly concerned with this. There is probably some legit reason why the system decided not to check the two devices in question. But it would still be nice to understand why. Might be time to read mdadm source code. | 12:53 |
onefang | Just a wild guess, since I never played with that, it's not checking those two .. yet? | 12:54 |
systemdlete | possibly. I thought of that also. But it still makes me wonder why | 12:54 |
onefang | Knowing rather than guessing is better. B-) | 12:55 |
systemdlete | agreed. Even if it turns out to be nothing | 12:55 |
systemdlete | I'm browsing the script called from cron. Considering the options it is passing, it should have checked ALL assembled arrays. I have 8 that are assembled. | 13:12 |
systemdlete | late now here. more tomorrow. | 13:16 |
systemdlete | bbl | 13:16 |
EHeM | In other news stable is still in a non-existant state. | 19:42 |
Evilham | Yup | 23:42 |
gnarface | you guys don't actually need it for something do you? | 23:53 |
gnarface | it's usually advised to use the release name in the sources.list anyway | 23:53 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!