aitor_ | there is something new in apt, apt-get install doesn't work without libudev.so.1, only dpkg | 00:01 |
---|---|---|
aitor_ | that's new for me | 00:01 |
aitor_ | today i've been able to build the sources of vdev in beowulf, but my images didn't boot neither in live mode nor after a hard disk installation | 00:03 |
aitor_ | if apt doesn't work without libudev.so.1, then it'll be quite complicate to change from udev to any other *dev and vice versa | 00:06 |
aitor_ | nobody here... ok, time to bed | 00:08 |
aitor_ | bye | 00:08 |
rwp | That is a new library for apt-get. Seems to have appeared between Stretch and Buster/Beowulf. | 01:19 |
fsmithred | apt has been forked in chimaera/ceres 2.1.11+devuan1 | 01:26 |
rwp | fsmithred, apt the single /usr/bin/apt program? Or apt the entire APT ecosystem of apt-get, aptitude, and everything? | 01:49 |
fsmithred | apt | 01:50 |
hemimaniac | ux | 01:51 |
fsmithred | apt policy apt or apt source apt | 01:51 |
Xenguy | fsmithred: Why has apt been forked? | 02:20 |
fsmithred | I think it was for libsystemd0 depends | 02:22 |
rwp | The upstream apt depends upon systemd? That does sound egregious in the extreme. | 02:24 |
fsmithred | just lsd0 or led0 | 02:27 |
fsmithred | but apt is important enough that someone decided to fork it | 02:27 |
Xenguy | thanks for the explanation fsmithred , this case really keeps it real | 02:35 |
Xenguy | apt | 02:35 |
mason | Xenguy: Part of the trouble is that libsystemd0 perpetuates the hairball. It's not a library with a coherent focus. | 02:35 |
Xenguy | rwp: "egregious" indeed | 02:35 |
mason | Xenguy: I wrote about this on DNG, but the trouble is that the authors want it to be another glibc, another universal library that everyone must link, and there's no reason for this beyond lazy design. But it has the effect of spreading the dependence. | 02:36 |
Xenguy | One wonders how much eventually needs to be forked : -/ | 02:37 |
mason | Xenguy: An increasing number of things. | 02:37 |
Xenguy | That's going to take volunteers | 02:37 |
mason | I need to brush up on it so I can start maintaining stuff I use that's slipped. | 02:38 |
Xenguy | Hopefully the community can rise to the challenge | 02:39 |
stovepipe | weird thing happened with slim not being able to start session and breaking itself, has anyone else seen this? | 02:42 |
stovepipe | it also happened to me on freebsd with fvwm | 02:43 |
stovepipe | it resolved itself and i'm not sure why, but this time on ascii | 02:43 |
stovepipe | i had to fiddle around and still not sure what eventually fixed it | 02:43 |
Xenguy | Running Ascii here, and no issues with 'slim' | 02:43 |
stovepipe | it seemed to leave skidmark lockfile in /tmp | 02:43 |
Xenguy | Maybe someone else can replicate the problem | 02:44 |
stovepipe | i think the alst thing i tried was shutting down slim, killing X, deleting the stuff in /tmp and restarting slim | 02:44 |
stovepipe | then after reboot it came up ok | 02:44 |
stovepipe | thats with xfce on ascii, on freebsd i'm using fvwm | 02:44 |
Xenguy | Nice to see fvwm is still around | 02:45 |
stovepipe | that one seemed to fix itself without me doing anything other then restarting slim or w/e a couple times | 02:45 |
stovepipe | both are set to autologin | 02:45 |
Xenguy | Sometimes stuff can shake down, and resettle, yay | 02:45 |
stovepipe | Xenguy: heh oh yeah, i've been evolving the same config for about 26 years now | 02:45 |
Xenguy | You are hardcore : -) | 02:46 |
stovepipe | yeah its a mystery as to what exactly happened or how i fixed it heh | 02:46 |
Xenguy | Which config is this? | 02:46 |
stovepipe | Xenguy: fvwm is just so perfectly tuned to what i want, why would i have it any other way | 02:46 |
stovepipe | Xenguy: i meant my fvwmrc | 02:47 |
Xenguy | Gotcha, I forgot for a second | 02:47 |
Xenguy | I think that was the first WM I ever tried | 02:47 |
stovepipe | anyways, if anyone gets stuck on that same slim problem | 02:47 |
Xenguy | I liked it, what can I say | 02:47 |
stovepipe | i first notices i couldnt change volume, because pulseaudio was down, because X wasnt starting, i dont actually use X on it its my server, and plays mpd 24/7 | 02:48 |
stovepipe | slim was starting and trying autologin and just sitting at a black screen | 02:48 |
stovepipe | on freebsd at that point it would drop back to the slim login | 02:48 |
Xenguy | Any idea what caused the problems? | 02:49 |
stovepipe | nope | 02:49 |
Xenguy | huh | 02:49 |
stovepipe | the only clues i got were the skidmark lockfiles on ascii | 02:49 |
stovepipe | on freebsd it magically fixed itself before i had a chance to look | 02:50 |
stovepipe | but both are slim autologin, so seems like the common element | 02:50 |
stovepipe | turning off autologin didnt change anything | 02:50 |
stovepipe | after several reboots after noticing the lockfile skidmarks, at one point i shutdown slim, deleted the skidmarks, restarted slim, then rebooted from xfce | 02:51 |
stovepipe | and it started working again | 02:51 |
stovepipe | if i'm remembering correctly heh | 02:52 |
Xenguy | All's well that ends well | 02:52 |
stovepipe | and a kill -9 X somewhere in there | 02:52 |
Xenguy | with prejudice! | 02:52 |
Xenguy | hah | 02:52 |
stovepipe | heh | 02:53 |
Xenguy | But seriously Xenguy_ , this *is* a support channel | 02:53 |
Xenguy | Xenguy_: Fine then, very well | 02:53 |
clort | runs fine for me | 04:20 |
bernte | I am getting Hash differences updateing beowulf Contents-amd64.tgz ... anybody else seeting the same? | 15:05 |
n4dir | someone reported such this morning (morning here), but i didn't read if a solution was found. Perhaps there is a log bout this IRC channel somewhere | 15:07 |
bernte | ok - just wanted to make sure I was not alone and PEBCAK ... | 15:11 |
onefang | Which package mirror are you using? | 15:11 |
bernte | E: Failed to fetch http://deb.devuan.org/merged/dists/beowulf/main/Contents-amd64.gz Hash Sum mismatch | 15:12 |
onefang | It's possible you tried just as it was updating. It's working fine for me right now. | 15:12 |
onefang | Ah the mirror DNS round robin. | 15:12 |
bernte | Which one are you using? | 15:12 |
onefang | http://sledjhamr.org/devuan/merged which is my own mirror. | 15:13 |
bernte | Ah - ok ... will stick to an offial one. | 15:13 |
onefang | My own mirror IS an official one. I'm the Devuan mirror herder. B-) | 15:14 |
onefang | https://pkgmaster.devuan.org/mirror_list.txt | 15:14 |
bernte | Oops .. :-) | 15:14 |
onefang | Did you wait half an hour and try it again? | 15:15 |
bernte | I will do that, though have tried it already for a while and also with DNS-RR this should not be so reliable. | 15:15 |
onefang | All the mirrors on the DNS-RR are passing my tests right now. | 15:17 |
bernte | is the /merged postfix correct? | 15:18 |
onefang | Yep. | 15:18 |
bernte | I consistently get SHA256:7fc4feca94409940e9e5d850fe2ebdc2837d82a848635e5cba4eb5e9881718c5 vs SHA256:79fa239a1624d16b62d271aee3c73b017d9a6bf099702d425cd453393cd21a2c | 15:19 |
bernte | bizzar - will continue digging, than your for confirming. | 15:19 |
bernte | s/than your/thank you/ | 15:20 |
gnarface | https://sledjhamr.org/apt-panopticon/results/Report-web.html | 15:21 |
gnarface | this was linked earlier | 15:21 |
gnarface | seems like there's still errors listed on some mirrors | 15:21 |
onefang | Though I need to patch up apt-panopticon a bit, m last change introduced some false errors. lol | 15:21 |
bernte | My error looks genuine: when I get https://sledjhamr.org/devuan/merged/dists/beowulf/InRelease Contents-amd64.gz has a Hash starting with 7fc4, when I wget the file and sha256, I get a Hash starting with 79fa .. *sigh* ... | 15:26 |
bernte | My sha256sum broken?! | 15:27 |
onefang | 79fa239a1624d16b62d271aee3c73b017d9a6bf099702d425cd453393cd21a2c Contents-amd64.gz | 15:30 |
onefang | That's on sledjhamr.org directly. | 15:31 |
onefang | 7fc4feca94409940e9e5d850fe2ebdc2837d82a848635e5cba4eb5e9881718c5 32744644 main/Contents-amd64.gz | 15:32 |
bernte | On on sledjhamr.org, the InRelease file say: 7fc4feca94409940e9e5d850fe2ebdc2837d82a848635e5cba4eb5e9881718c5 32744644 main/Contents-amd64.gz | 15:32 |
onefang | That's what I was about to say. lol | 15:32 |
bernte | Ok - at least we agree what we see :-) | 15:33 |
onefang | ccc389a3af882e45e953540760b200328c235219d1cf7e2549e70a02591eef62 32742184 main/Contents-amd64.gz | 15:34 |
onefang | That's on pkgmaster, which is what all the mirrors sync from. | 15:34 |
onefang | AND the actual sha256 of that file agrees. | 15:35 |
onefang | Oh wait, ignore that, I was looking at an old backup on pkgmaster. lol | 15:36 |
bernte | :-) | 15:36 |
onefang | 7fc4feca94409940e9e5d850fe2ebdc2837d82a848635e5cba4eb5e9881718c5 32744644 main/Contents-amd64.gz | 15:37 |
onefang | 79fa239a1624d16b62d271aee3c73b017d9a6bf099702d425cd453393cd21a2c Contents-amd64.gz | 15:37 |
onefang | So the source that all the package mirrors sync to has the same problem. | 15:38 |
onefang | I'll escalate this to the people responsible for upstream from there. | 15:38 |
bernte | ok - thank you. I will just wait until tomorrow and see if the issue persists. | 15:39 |
tuxd3v | in chimaera each time I lock my screen | 19:52 |
tuxd3v | it disconect the irc | 19:52 |
tuxd3v | i am using hexchat | 19:52 |
tuxd3v | this behaviour doesn't hapen in beowulf | 19:53 |
systemdlete | rwp: Why is something like pam_systemd.so even in devuan in the first place? Did this get overlooked during surgery? | 20:03 |
bru | tuxd3v: With what software do you lock the screen? Or do you flip down the laptop screen? | 20:04 |
tuxd3v | I am testing Mate Desktop | 20:25 |
tuxd3v | so I am using mate one | 20:25 |
tuxd3v | in Beowulf I am using slim, which has slomlock | 20:26 |
tuxd3v | err, slomlock -> slimlock | 20:26 |
Hurgotron | Beowulf install on Lenovo T500, keyboard doesn't work. Any hints? | 21:19 |
fsmithred | Hurgotron, does the trackpad work? | 21:26 |
fsmithred | try plugging and unpluggin a usb mouse or keyboard and see if it wakes up the laptop keyboard. | 21:34 |
Hurgotron | it did... | 21:48 |
ranix | wtf | 21:48 |
Hurgotron | how can I make sure it'll be usable without an extra keyboard...? | 21:49 |
ranix | that was guru level fsmithred | 21:49 |
systemdlete | crontab -l is broken on one of my ascii systems. Tells me fopen: permission denied. Found a few so-called solutions online, but none of them seem applicable. One suggested directory perms on /var/lib/cache/crontabs, but the perms were OK. So I am not sure what the problem is. | 21:50 |
systemdlete | I *am* seeing a lot of errors in the daemon.log about consolekit failing... every 5 seconds! | 21:50 |
fsmithred | back | 21:51 |
fsmithred | Hurgotron, get eudev from beowulf-proposed-updates | 21:51 |
fsmithred | or... add acpi-fakekey (I think that'll do it) | 21:51 |
fsmithred | systemdlete, did you try rebooting? | 21:52 |
systemdlete | fsmithred: Yes. In fact, I did update/upgrade too before reboot | 21:52 |
fsmithred | I've occasionally gotten some weird permission errors | 21:53 |
fsmithred | hm | 21:53 |
systemdlete | Seems to be related to control groups. | 21:53 |
systemdlete | ring a bell? | 21:53 |
fsmithred | nope | 21:53 |
systemdlete | well, just thought I'd throw that one out there | 21:53 |
fsmithred | I never chased down the cause of the problem | 21:53 |
fsmithred | sometimes nano won't open for me | 21:53 |
systemdlete | I'm in the middle of migrating to a beowulf system, but it is not ready yet. | 21:53 |
systemdlete | Moments like this are when I want to go find a certain person/corporation and do what they do to chickens before they cook the chickens. | 21:54 |
ranix | rontab -e creates a temporary file for editing and then copies the resulting updated file to the crontab when the spawned editor is closed I believe | 21:54 |
systemdlete | They broke our necks, so it seems fair. | 21:54 |
ranix | so it could be having a permission error reading the original crontab, making the temporary file, making a temp folder, or writing to the temp folder | 21:55 |
systemdlete | ranix: Good idea! I could run strace and/or ltrace and see what it is *trying* to do | 21:55 |
systemdlete | although I may run into other perms problems due to the need to escalate privileges, idk. | 21:56 |
systemdlete | bbl | 21:57 |
Hurgotron | The dektop scaling seems to be totally frong too, everything is twice as wide... | 22:00 |
systemdlete | ok, one thing I see is that fopen is failing to open /var/spool/cron/crontabs/myusername/ which would be a directory, but it does exist as a file. | 22:02 |
fsmithred | can you fix that with the display settings? | 22:02 |
Hurgotron | I seem to have a virtual desktop vor whatever reasons, I can scroll it left and right | 22:03 |
fsmithred | weird | 22:03 |
fsmithred | xfce? | 22:03 |
ranix | systemdlete: sounds like you found the problem or close enough to it to hack in a forced workaround | 22:04 |
fsmithred | Hurgotron, does audio work? | 22:04 |
ranix | someone probably added support for crontab.d style directories or something and it's choking on that because your system doesn't match the assumptions | 22:05 |
Hurgotron | fsmithred: No | 22:06 |
systemdlete | I'm wondering, though, why it would be trying to fopen on what appears to be a directory path (trailing slash). | 22:06 |
systemdlete | Linux distros used to work so nicely... | 22:06 |
systemdlete | Then they had to go and "improve" them. | 22:06 |
systemdlete | So this. | 22:06 |
mason | systemdlete: That should be a file, not a directory. | 22:06 |
systemdlete | mason: I agree. | 22:06 |
systemdlete | So wtf is going on? | 22:07 |
fsmithred | Hurgotron, you need to fix eudev. Either get the newer version or add 'sleep 1' to /etc/init.d/eudev | 22:07 |
fsmithred | I'll tell you where | 22:07 |
mason | systemdlete: What is there - is it a directory? I can argue either way based on wording. | 22:07 |
systemdlete | (I just saw a RedHat commercial on youtube, for the first time, today) | 22:08 |
systemdlete | mason: It is a file, as it should be. But the fopen error message (as well as the strace) shows it trying to fopen the spec'd path as a directory. | 22:08 |
systemdlete | (and when I saw it, my eyes bled) | 22:09 |
mason | systemdlete: Can you repeat the error message? fopen generally just wants a file, whatever it is. | 22:09 |
systemdlete | (but that's O.T. here...) | 22:09 |
systemdlete | Yes, mason. It is entirely reproducible. | 22:09 |
mason | systemdlete: And can you link me the ad? Curious if I've seen it. | 22:09 |
systemdlete | Every time I run "crontab -l" as my non-root user. | 22:10 |
systemdlete | mason: Was watching youtube on my TV, so... not sure how to get that for you. | 22:10 |
mason | oh | 22:10 |
systemdlete | sorry, man | 22:10 |
systemdlete | would love to though | 22:10 |
mason | no worries - so, what's the error message? | 22:10 |
systemdlete | share the pain | 22:10 |
ranix | your non-root user has a crontab? | 22:11 |
mason | ranix: Any user can have one. | 22:11 |
systemdlete | crontabs/myuser/: fopen: Permission denied | 22:12 |
mason | systemdlete: And are you use your crontab isn't, say, a symlink to something? | 22:12 |
systemdlete | nope. Just a plain file | 22:12 |
systemdlete | It has 0600 perms, owned by cron, group myuser | 22:13 |
mason | systemdlete: Can you share ls -laZ /var/spool/cron/crontabs/systemdlete or whatever? Change the name if you want. | 22:13 |
systemdlete | I tried changing it to 660, but no help | 22:13 |
systemdlete | oh... didn't think of that! | 22:13 |
ranix | what happens if you move it to /tmp | 22:13 |
ranix | and crontab -e | 22:13 |
mason | systemdlete: Check the directory permissions too. | 22:13 |
systemdlete | it shows question mark | 22:14 |
ranix | also can you cat the original /var/spool/cron/crontabs/myuser | 22:14 |
mason | systemdlete: Check directory permissions for all of /var, /var/spool, /var/spool/cron, etc. | 22:14 |
systemdlete | ranix: I can manipulate the same file as root | 22:14 |
ranix | ls -hal /var/spool/cron/crontabs | 22:14 |
ranix | move or delete it | 22:14 |
ranix | then crontab -e as the user | 22:14 |
ranix | does that work | 22:14 |
mason | systemdlete: https://bpa.st/OPPQ | 22:15 |
ranix | crontrab doesn't have r access to /var/spool/cron/crontabs? | 22:15 |
mason | Doesn't need it. | 22:16 |
mason | Only needs execute. | 22:16 |
Hurgotron | fsmithred: eudev from beowulf-proposed-updates helped. Thanks a lot | 22:16 |
ranix | yeah looks like those perms are the same on mine | 22:16 |
ranix | yeah looks like those perms are the same on my machine | 22:16 |
fsmithred | yw | 22:16 |
Hurgotron | (my nerves are not the best currently) | 22:16 |
ranix | maybe you have some weird settings in /etc/default/cron | 22:17 |
systemdlete | OK, "myuser" is able to cd as far as /var/spool/cron | 22:18 |
mason | ranix: It's almost certain he has bad directory permissions along the path. | 22:18 |
systemdlete | that user cannot cd into crontabs subdir | 22:18 |
ranix | that's normal | 22:18 |
systemdlete | right | 22:18 |
systemdlete | and I tried moving the file but no luck either | 22:18 |
mason | systemdlete: Are you ignoring what I'm telling you intentionally? | 22:18 |
systemdlete | it failed this time with fdopen, not fopen | 22:18 |
systemdlete | mason: No | 22:18 |
systemdlete | but both of you are asking me questions and things to try | 22:19 |
ranix | he also hasn't tried moving the file then making a new one as the user | 22:19 |
systemdlete | ranix: Yes I did! | 22:19 |
systemdlete | ^^^ | 22:19 |
ranix | oh, did it work | 22:19 |
systemdlete | (read) | 22:19 |
mason | Well, I'll bow out. ranix has it in hand. | 22:19 |
ranix | idk if I have it in hand, he's telling me to scroll up to find where he answered my suggestion from awhile ago in the scrollback | 22:20 |
systemdlete | mason: Yes, the perms are identical to yours. | 22:20 |
systemdlete | ranix: I can see my responses to you. | 22:21 |
systemdlete | you should not need to "scroll up" | 22:21 |
systemdlete | my replies are in the last 2 minutes! | 22:21 |
systemdlete | maybe there is a problem with your irc client? | 22:21 |
ranix | ok, I'm gonna bow out and not play 100 questions with you asking where exactly it is | 22:21 |
ranix | I assume it's the thing you posted about fdopen which is not clear | 22:22 |
systemdlete | <systemdlete> and I tried moving the file but no luck either | 22:22 |
systemdlete | <mason> systemdlete: Are you ignoring what I'm telling you intentionally? | 22:22 |
systemdlete | <systemdlete> it failed this time with fdopen, not fopen | 22:22 |
ranix | that's a lot of information for you to use to solve your problem that you generated during that test | 22:22 |
systemdlete | the game is called "20 questions," btw. | 22:22 |
mason | systemdlete: going all the way back: https://bpa.st/GISA | 22:23 |
mason | Also, /usr/bin/crontab ought to be: -rwxr-sr-x 1 root crontab 43K Oct 11 2019 /usr/bin/crontab | 22:23 |
systemdlete | mason: Yes, all of those directories in the path match yours | 22:24 |
mason | cahrm | 22:24 |
systemdlete | mason: Ah, that's a thought. Let me double-check that | 22:24 |
mason | hrm* | 22:24 |
mason | crontabs itself can execute, global cannot, user cannot on his own | 22:25 |
systemdlete | you dunnit, mason! THe smoking gun | 22:25 |
mason | \o/ | 22:25 |
mason | Sorry I didn't think of that earlier. | 22:25 |
systemdlete | Mine is root/root with 755 | 22:25 |
systemdlete | no setuid | 22:25 |
mason | It's going to give the same error as if the directory permissions were wrong. | 22:25 |
systemdlete | right | 22:25 |
mason | well, setgid | 22:25 |
systemdlete | right, and change group too | 22:25 |
systemdlete | thanks, buddy | 22:26 |
mason | so, move 2755, chgrp crontab | 22:26 |
systemdlete | Sorry YOU didn't think of that earlier? What about ME? | 22:26 |
systemdlete | right | 22:26 |
mason | Eh, you're here asking for help. If I'm offering it I should be thinking of the things. | 22:26 |
mason | mode* | 22:26 |
systemdlete | And, honestly, I was not ignoring you at all. It's a different box, so it took me some time to try things and respond to you. | 22:28 |
systemdlete | Sorry about that. | 22:28 |
systemdlete | well, now it is root/crontab with 2755 perms. Still, "myuser" cannot do crontab -l -- same fopen error. | 22:29 |
systemdlete | Not sure how the program got changed, either. | 22:29 |
mason | systemdlete: That's alright. My hubris is its own punishment. I was insisting that you look at incorrect answers. | 22:29 |
systemdlete | I wonder if I should purge and re-install crontab | 22:29 |
systemdlete | mason: No need to feel bad, really. | 22:30 |
mason | systemdlete: I'd look for other deviations. I'd be worried about that being the result of a global (or directorywide) clobber of some sort. | 22:30 |
mason | systemdlete: Not bad, but it's a lesson for me. :) | 22:30 |
fsmithred | how did it get that way? | 22:30 |
systemdlete | no idea. | 22:30 |
fsmithred | not something you did? | 22:30 |
systemdlete | I can't say it isn't, tbh. I just don't remember changing anything at the level of system-wide modifications in /etc | 22:31 |
systemdlete | But it is possible. | 22:31 |
systemdlete | The system continues to work, but I am migrating to beowulf. I don't want to spend too much on this, but it is weird. | 22:31 |
systemdlete | oh. | 22:31 |
systemdlete | doh! | 22:31 |
systemdlete | I stopped cron to try ranix's experiments, and forgot to turn it back on. | 22:32 |
systemdlete | shit | 22:32 |
systemdlete | restarted cron but still no luck | 22:34 |
mason | systemdlete: Is it not picking up what you've got in your crontab? | 22:34 |
mason | systemdlete: Verify the crontab permissions. Example: -rw------- 1 mason crontab 1169 Nov 13 22:58 mason | 22:34 |
systemdlete | good question. This had been working months ago. | 22:34 |
mason | systemdlete: And a good test: * * * * * /bin/touch /tmp/crontest | 22:35 |
systemdlete | mason: I have an entry already that runs every minute... there might be a log | 22:35 |
mason | The fewer the moving parts, the better the test, but either was ought to give you information, sure. | 22:36 |
mason | either way* if I could type | 22:36 |
systemdlete | I tried your test mason. And that worked (I shut down cron, modified the file, restarted cron) | 22:38 |
systemdlete | (to avoid crashing cron or causing sync problems for it) | 22:38 |
mason | BTW, you shouldn't have to restart cron. It will notice changed files. | 22:38 |
systemdlete | If root is in the middle of updating the file? | 22:38 |
mason | systemdlete: A quick note is that sometimes container filesystems defeat cron's default method of checking for changes. | 22:39 |
systemdlete | this is a vbox VM. But thanks for the tip. | 22:39 |
systemdlete | mason: Any ideas on all those consolekit errors in daemon.log? They go off every 5 seconds. | 22:40 |
systemdlete | I'm wondering if that is connected at all. | 22:41 |
mason | Can you share an example, or did you earlier? | 22:42 |
systemdlete | http://paste.debian.net/1177876/ | 22:42 |
systemdlete | (was preparing it just as you were asking...) | 22:43 |
mason | systemdlete: Ah, I'd suspect that's going to be a similar issue with a bad permission somewhere. Hrm. | 22:44 |
systemdlete | I agree. | 22:44 |
systemdlete | maybe consolekit has a trace mode | 22:45 |
mason | systemdlete: That looks like a systemd-specific issue, as it turns out. | 22:47 |
systemdlete | Oh, definitely. I already figured that. LOL! | 22:47 |
systemdlete | The surgeon might have left a scalpel or clamp before he closed up. | 22:48 |
mason | checking | 22:48 |
systemdlete | I'm getting very used to this sort of thing. | 22:48 |
systemdlete | mason: Consolekit has a debug flag! | 22:50 |
mason | systemdlete: That.... appears to be a RHEL-specific error even. I don't know why it's showing up on a Debian-derivative at all. | 22:51 |
systemdlete | mason: I'm going to try the --debug option on consolekit. Give me a few minutes to try this. | 22:53 |
mason | Oh, I might be barking up the wrong tree again. Backing up, starting again. | 22:53 |
mason | kk | 22:53 |
systemdlete | huh. | 22:54 |
systemdlete | I don't even see consolekit running or anywhere in /etc/init.d... I must need new eyeglasses. | 22:54 |
mason | Maybe that's the issue...? | 22:55 |
systemdlete | nah. I think it is dbus | 22:55 |
systemdlete | dbus is the one trying to call consolekit or whatever it is doing... | 22:55 |
mason | I'm horrible at debugging dbus, in part because I really dislike it. | 22:56 |
mason | Vicious circle. | 22:56 |
systemdlete | a lot of people do | 22:56 |
systemdlete | I wish people would stop trying to fix things that weren't broken. | 22:56 |
systemdlete | Or, at least, make their changes backward-compatible. | 22:56 |
systemdlete | dbus, sadly, has no built-in debug or trace | 23:03 |
mason | systemdlete: you can strace it | 23:04 |
systemdlete | true that | 23:04 |
mason | systemdlete: strace -ftttTvyyo /tmp/dbus.trace -s 4096 -p <dbus PID> | 23:05 |
systemdlete | Well, I tried that mason. There are 4 processes running as varying invocations of dbus. One of them will not allow attachment by strace. The other 3 all spit out one message and sit there. | 23:10 |
systemdlete | strace flush option? | 23:10 |
mason | systemdlete: You should be able to attach as root. | 23:10 |
systemdlete | ah good thx | 23:11 |
systemdlete | Yes. Now the errors are illuminated somewhat. "writev(2</dev/null>, [{iov_base="dbus[2118]: [system] Activated service 'org.freedesktop.ConsoleKit' failed: Failed to execute program org.freedesktop.ConsoleKit: Permission denied", iov_len=147}, {iov_base="\n", iov_len=1}], 2) = 148 <0.000061>" | 23:12 |
systemdlete | oh, wait. Look at this: | 23:14 |
systemdlete | [pid 3629] 1608588742.082334 open("/proc/self/oom_score_adj", O_WRONLY) = -1 EACCES (Permission denied) <0.000040> | 23:14 |
systemdlete | [pid 3629] 1608588742.082451 fcntl(-1, F_GETFD) = -1 EBADF (Bad file descriptor) <0.000028> | 23:14 |
systemdlete | This comes before the other error message | 23:14 |
mason | Hrm, but if that's the one running as root, .... odd | 23:14 |
systemdlete | well, hold on. Let me check | 23:15 |
systemdlete | pid 2118 was the one running as root, the output of which I am looking at | 23:15 |
systemdlete | looks like that process (3629) was ephemeral. It's gone now. So it must have been something else. | 23:16 |
mason | Probably worth verify dbus binary permissions or reinstalling some/all of it. | 23:16 |
systemdlete | (I did some copy/paste to show you output; maybe related to that, idk) | 23:16 |
systemdlete | mason: Recall that this VM is going away. I don't want to put that much work into things. The system works the way I wanted it to. It's just that crontab -e/-l no longer work for regular user for some reason. Sorry if I was unclear. | 23:17 |
systemdlete | And this IS ascii, not beowulf. | 23:17 |
systemdlete | So if you don't want to put any more effort in on this, I understand. | 23:17 |
systemdlete | (unless you are very curious) | 23:18 |
systemdlete | I do not want to waste your time or patience. | 23:18 |
systemdlete | dbus is running and seems to be spitting out lots of errors. | 23:18 |
systemdlete | mason: dbus has 755 perms and is root/root owned | 23:19 |
systemdlete | same as on this beowulf system I am using to chat with you | 23:19 |
mason | Hm, not seeing anything unusual here for permissions of /usr/bin/*dbus* | 23:20 |
systemdlete | Maybe I should just resign myself to having to work around this (use root to modify the user's crontab) until I'm done migrating this VM. | 23:20 |
systemdlete | mason: What are the perms on your *dbus* binaries please? I'll compare mine | 23:23 |
systemdlete | mine are all 755 root/root | 23:23 |
mason | systemdlete: same: https://bpa.st/44FQ | 23:24 |
systemdlete | mason: Let's say I were to re-install dbus completely (purge/install?). Would that require complete reinstall of MATE? | 23:24 |
systemdlete | I mean --reinstall would work, but if any of dbus's configs are involved in this... | 23:25 |
mason | systemdlete: I'd think you could force-reinstall. | 23:26 |
systemdlete | does that purge the config also? | 23:26 |
mason | systemdlete: man dpkg, look at the various --force-things options | 23:27 |
systemdlete | ok | 23:27 |
mason | in this case, look in particular at confnew | 23:27 |
fsmithred | if you're not sure what configs were changed, install and run 'debsums -ca' | 23:27 |
systemdlete | fsmithred: Could I run debsums -ca right now and tell which configs changed from the template/defaults? | 23:29 |
fsmithred | yeah, if it's installed | 23:29 |
systemdlete | oh dbus is installed, alright... | 23:30 |
systemdlete | so debsums is like rpm's audit | 23:31 |
systemdlete | nice to know, thanks | 23:31 |
systemdlete | well, as it turns out, nothing that jumps out at me. All of the changes debsums listed I know what those were for; there were no surprises there. | 23:33 |
systemdlete | Most of them were for my backup system. The only one that might be suspect -- a little perhaps -- would be sysctl.conf | 23:33 |
systemdlete | And I know I've modified that a few times for various things. | 23:33 |
systemdlete | Wish debsums had an option to give diff output | 23:40 |
systemdlete | (didn't see that in man page) | 23:40 |
tuxd3v | bru, I believe I found the culprit :) | 23:45 |
tuxd3v | I had the machine to sleep after 30 min.. | 23:46 |
tuxd3v | Now I will test with 'never', so that my sessions mainstain themselfs up :) | 23:46 |
tuxd3v | themselves* | 23:46 |
systemdlete | mason, fsmithred: So maybe "apt-get install --reinstall --force-confnew dbus" ? | 23:47 |
systemdlete | (it tells me there's a bad combo of options) | 23:48 |
systemdlete | or maybe more like "dpkg --force-reinstall --force-confnew dbus" ? | 23:50 |
mason | systemdlete: the force-confnew is dpkg | 23:55 |
systemdlete | mason: How do I also tell dpkg that I want to force a complete reinstall of just that package (dbus) | 23:58 |
systemdlete | ? | 23:58 |
mason | systemdlete: I believe what you want is to fetch the package and dpkg -i with the selected options | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!