rwp | I think memory leaks are definitely on topic. | 00:02 |
---|---|---|
rwp | I am not using Mate (or other DE at all) but Firefox is definitely a memory pig. | 00:03 |
rwp | However the Internet web today pretty much consumes all resources to operate as designed by the web site designers. | 00:03 |
rwp | A clock however should reach steady state resource use pretty quick and then stay there. | 00:04 |
bigdumbelk | any of you all install devuan on a librem 15 | 00:22 |
fsmithred | systemdlete, is 1.22 new enough? | 00:22 |
fsmithred | http://hezeh.org/mate-desktop-1.22/ | 00:23 |
bigdumbelk | want to get a systemd free system with kill switches in a librem 15 | 00:23 |
systemdlete | fsmithred: I'm not sure what standard the mate devs go by. They say the current release is 1.24. They did not mention anything re 1.22. | 00:55 |
fsmithred | hezeh.org has 1.22 for beowulf | 00:56 |
systemdlete | fsmithred: Devuan Beowulf is not a modern distro... uhm. | 00:56 |
systemdlete | lol | 00:56 |
fsmithred | it's current stable based on debian's current stable. | 00:57 |
systemdlete | yep. | 00:57 |
fsmithred | So of course it's not new. | 00:57 |
fsmithred | chimaera and ceres have 1.24 | 00:57 |
systemdlete | If I follow the argument literally, then I guess MATE requires whatever-is-the-tip-version-in-the-source-repo | 00:58 |
fsmithred | or read the changelogs to see where they fixed the panel | 00:58 |
systemdlete | Maybe the REAL solution is not to use MATE | 00:58 |
fsmithred | xfce panel works in beowulf and chimaera/ceres | 00:58 |
systemdlete | I think the issue may have to do with libraries that changed in debian 11 | 00:59 |
fsmithred | pretty sure lxpanel works, too | 00:59 |
fsmithred | check anto's version too. (hezeh) It might be fixed there. He backports mate and cinnamon for devuan. | 01:00 |
systemdlete | yes, I have had decent luck with xfce nearly everywhere. lxde and lxqt have been somewhat less reliable. Sometimes I found they didn't have all the pieces | 01:00 |
systemdlete | ok, will do. | 01:00 |
systemdlete | fsmithred: Is the fact I installed using star not devuan matter? I'm thinking that's mainly installer choice. They all use the same repos, right? | 01:01 |
fsmithred | afaik | 01:01 |
fsmithred | look in sources.list | 01:01 |
fsmithred | yeah, I do know. Star is just devuan. | 01:02 |
systemdlete | they are the same. | 01:02 |
systemdlete | I'll give hezeh a go | 01:02 |
systemdlete | thanks for your patience again fsmithred! | 01:03 |
fsmithred | dpkg -l task-mate-desktop | 01:03 |
fsmithred | see if the metapackage is installed | 01:03 |
systemdlete | dpkg-query: no packages found matching task-mate-desktop | 01:03 |
systemdlete | I think I might have installed it differently, but idr now | 01:04 |
fsmithred | cool | 01:04 |
fsmithred | that'll make it easier to remove stuff | 01:04 |
systemdlete | antofox-keyrings is missing? | 01:09 |
systemdlete | do I need to contact the dev? | 01:09 |
systemdlete | apt-get install antofox-keyrings --allow-unauthenticated | 01:09 |
systemdlete | (instructions on page at hezeh.org) | 01:09 |
systemdlete | btw, there is no www.hezeh.org, only hezeh.org, it seems | 01:10 |
systemdlete | other than that, the update worked, but gave me the error about needing the key. So I did that and got this error | 01:10 |
fsmithred | add the key and then apt update again | 01:10 |
systemdlete | can I use the keyserver instead of his instructions? | 01:11 |
fsmithred | use what keyserver? | 01:11 |
systemdlete | keys.gnupg.net | 01:11 |
fsmithred | I think I downloaded it from his site directly | 01:12 |
systemdlete | well, the instructions at his page are not working for me. I'll try adding the key with the method I know | 01:12 |
fsmithred | http://hezeh.org/packages/pool/main/a/antofox-keyrings/ | 01:13 |
fsmithred | download the .deb and install with dpkg | 01:13 |
systemdlete | that worked to fix the key issue. Thanks. | 01:19 |
systemdlete | But following his instructions doesn't seem to work. Yours do. | 01:19 |
systemdlete | he wants me to install mate-desktop-environment, but after the apt update, apt already knows I want to update all the mate packages. | 01:20 |
systemdlete | Can I just do the update, and skip the mate-desktop-environment? | 01:20 |
fsmithred | probably | 01:21 |
fsmithred | is that package already installed? | 01:21 |
systemdlete | I see that apt wants to install all the 1.22 packages anyway | 01:21 |
systemdlete | not sure | 01:21 |
fsmithred | dpkg -l |grep mate | 01:21 |
systemdlete | yes. It says it already has latest version | 01:21 |
fsmithred | will show you all the installed packages with mate in the name | 01:21 |
fsmithred | gotta go. bbl. | 01:22 |
systemdlete | fsmithred: I upgraded one of my 2 star VM's and running my memory leak script for a few hours | 02:40 |
systemdlete | and re-enabled the "show seconds" just to be sure that works right now. | 02:43 |
systemdlete | and I am seeing 132 byte increases in memory by the panel again. | 02:43 |
Xenguy | systemdlete: How do you typically monitor the memory leakage in MATE. I confess I haven't been able to detect any memory problems with MATE, but I'm only looking on Ascii so far | 02:44 |
systemdlete | Looks like about once every 1m 20s | 02:44 |
Xenguy | er, leakage in mate-panel | 02:44 |
systemdlete | Xenguy: I have a memory leak script I wrote myself. | 02:44 |
Xenguy | huh, ok | 02:44 |
systemdlete | Xenguy: The leak is very tiny, so it is mostly unnoticeable. Trouble is, over a day or so, that can amount to several 100MBs of growth. After a few days or maybe weeks, it creates an OOM condition and all heck follows | 02:46 |
Xenguy | OK, will keep an eye on it as best I can, but so far so good | 02:47 |
systemdlete | (it could just be something local here, though so many others have reported it also, so it is not likely) | 02:50 |
systemdlete | Interesting to note that both mate-panel AND the clock applet are increasing by these 132 byte increments, though the clock not as frequently. | 02:58 |
systemdlete | More interesting yet is that the increases are not coming on the minute, but at varying times (number of seconds after the minute). | 02:59 |
systemdlete | So one has to wonder how/why these tiny memory increases are occurring. | 02:59 |
mason | systemdlete: Bad programming. | 03:02 |
systemdlete | mason: y'think? lol. yeah | 03:05 |
n4dir | did you shellcheck it? | 03:06 |
n4dir | if it is bash or such. didn't fully understand it. But sure ain't hard to screw such | 03:07 |
mason | systemdlete: It can be hard to track that sort of thing. Might be an actual memory leak, or it could be bookkeeping without realistic self-awareness. | 03:07 |
systemdlete | what does that mean? | 03:10 |
systemdlete | I would hope that frameworks for building things like desktop environments would offer some sort of memory use profiling. If nothing else, aren't there linters out there to do that sort of thing? | 03:10 |
systemdlete | n4dir: This is a perl script, and it is not the target we are looking at. My script scans ps output every so many seconds | 03:12 |
systemdlete | the target is mate-panel, and this is a known problem up until at least 1.24 (the latest according to mate devs) | 03:13 |
systemdlete | mate-panel is part of the MATE DE | 03:13 |
n4dir | get rid of mate panel then? | 03:13 |
systemdlete | Or just get rid of computers altogether why don't we. That is CERTAIN to eliminate all bugs and security problems, right? | 03:14 |
systemdlete | :p | 03:14 |
n4dir | you sure can use different panels or none at all in other DE's. iirc | 03:14 |
systemdlete | In that case, I'd rather just turn to a different DE altogether. But this is an interesting issue, and I kinda like MATE, ok? | 03:15 |
n4dir | and why investigate a known problem? | 03:15 |
n4dir | to me Mate didn't look outstanding ressource friendly in general | 03:16 |
mason | systemdlete: What I mean is, it might not be a memory leak in the sense of an error. It might be bad design insofar as it doesn't consider memory over time for little book-keeping details. I don't know - I don't use the stuff. | 03:17 |
systemdlete | I still don't follow. A memory leak is caused by not ensuring that memory is released once it is no longer needed. | 03:49 |
systemdlete | Or are you saying that the panel needs 132 more bytes of memory every few minutes because it is keeping a log of some sort internally? | 03:50 |
systemdlete | I agree with you that there is a basic design and/or implementation problem with it. | 03:57 |
systemdlete | n4dir: We were not investigating a known problem per se. We were investigating whether the 1.22 version had the corrections their devs referred to. It was worth a try. | 03:58 |
clort | sorry i missed your question n4dir | 04:08 |
clort | new youtube-dl works for me | 04:09 |
n4dir | i could give you the link, if you need: Kunst in Aktion. It is quite ok | 04:13 |
mason | systemdlete: That's the possibility, yeah, it's tracking something legitimately and not ever considering the impact over time. But I don't know. It's one of the possibilities. | 04:45 |
postal | hi guys how to fix this error. tried everything, nothing helps https://0bin.net/paste/af9BIbep#uXUv5kKSnrMI58qDJ5AUX3hJizJMNAt9UzKG7bxjhbY | 13:01 |
gnarface | postal: use paste.debian.net and i'll look at it | 13:02 |
clort | hi postal | 13:07 |
clort | can you describe the problem | 13:07 |
postal | gnarface http://paste.debian.net/1177700 | 13:16 |
postal | I can't apt upgrade for this error "Hash Sum mismatch" | 13:17 |
gnarface | postal: well you have deb-multimedia in there, which you shouldn't do and may also be causing problems, but my best guess is that you started this when "stable" was different, and now it's hosed because they changed which release is called "stable" out from underneath you | 13:19 |
gnarface | postal: use the proper release names and get deb-multimedia crap out of there and it should be able to complete the upgrade with a little massaging | 13:20 |
gnarface | postal: (you might have to manually remove some packages, finish updating from the old release, then complete an upgrade to the new release) | 13:20 |
fsmithred | could be that the mirror is in the middle of updating. Try again in an hour or two. | 13:21 |
fsmithred | using third-party repos is a crap shoot and using "stable" instead of the current codename "beowulf" may cause problems when the debian stable changes and devuan does not. | 13:22 |
gnarface | yea it might be relevant how long it has been since this started happening but i guess in my head i assumed "tried everything" included waiting a day | 13:26 |
gnarface | same error occasionally comes out of my proxy when the server updates between the proxy cache refresh interval | 13:27 |
gnarface | (clears itself within 5 minutes usually, or if i manually run an update without it) | 13:27 |
postal | still doesn't work even with current codename "beowulf" instead of "stable" and without deb-multimedia and other repos in sources.list.d http://paste.debian.net/1177701 | 13:42 |
gnarface | postal: did you try it once with ascii first? | 13:43 |
gnarface | postal: ascii instead of beowulf | 13:43 |
postal | I installed clean beowulf, although I don't remember much, since this is not my laptop, but my mother's, I just update it from time to time | 13:53 |
gnarface | oh, it's not an update from a previous version? | 13:55 |
gnarface | could you paste your /etc/apt/sources.list and the contents of anything in /etc/apt/sources.list.d/ ? | 13:55 |
gnarface | i'll just sanity check it for you | 13:55 |
gnarface | or, you could just remove it all and use this example instead: https://paste.debian.net/1177702/ | 14:02 |
postal | My sources.list and sources.list.d repos https://paste.debian.net/1177703/ | 14:17 |
gnarface | postal: and you're not running an apt proxy, right? then i'd say yea probably just wait and speak up again if it's not cleared in a few hours | 14:21 |
gnarface | usually doesn't need more than a half hour or so but i think it can regularly take 2? | 14:22 |
gnarface | don't quote me on that | 14:22 |
postal | i don't use apt proxy | 14:28 |
gnarface | just making sure | 14:29 |
fsmithred | https://sledjhamr.org/apt-panopticon/results/Report-web.html | 14:29 |
fsmithred | some mirrors are down | 14:29 |
fsmithred | postal, here's a possible fix for the problem: https://dev1galaxy.org/viewtopic.php?pid=11331#p11331 | 14:33 |
postal | unfortunately i do "rm -rf /var/lib/apt/lists/*" before apt update constantly, but doesn't help. | 14:48 |
captainfixerpc14 | model t car drawing | 17:57 |
captainfixerpc14 | oops | 17:57 |
fsmithred | can it run linux? | 18:01 |
captainfixerpc14 | working on that :P | 18:05 |
captainfixerpc14 | lsb_release -a | 18:14 |
captainfixerpc14 | arggggggg | 18:14 |
captainfixerpc14 | too many keyboards | 18:14 |
captainfixerpc14 | sorry | 18:15 |
systemdlete | lightdm: PAM adding faulty module: pam_systemd.so | 20:03 |
systemdlete | lightdm: PAM unable to dlopen(pam_systemd.so): /lib/security/pam_systemd.so: cannot open shared object file: No such file or directory | 20:03 |
systemdlete | hehehheh... there is no escaping it, is there? | 20:03 |
systemdlete | (it's EVERYwhere) | 20:04 |
systemdlete | fsmithred: Thanks for the suggestion of the mate 1.22 repo. Sadly, though, it did not help. Apparently the fixes are only in 1.24+. Still, thanks for trying. | 20:05 |
fsmithred | yw | 20:06 |
systemdlete | I could try filing for a backport, but since this is a library dependency problem, I doubt there will be much enthusiasm to do such. Better for me to wait for chimaera I think. | 20:06 |
fsmithred | how critical is this system? You might do ok with chimaera or chimaera/ceres | 20:07 |
systemdlete | fsmithred: Probably NOTHING I *ever* do is really "critical" per se. Mostly my own stuff. But I really detest instability. I'm funny like that. | 20:07 |
fsmithred | people are reporting very stable systems (I'm one of the people) | 20:08 |
fsmithred | I'm running chimaera xfce on an old thinkpad. Don't do a whole lot with it, but it's working. | 20:08 |
fsmithred | I do jitsi with it, I ssh into it to check stuff 'cause I'm too lazy to turn around. | 20:09 |
systemdlete | So devuan project is waiting on debian to release theirs before devuan releases ours? | 20:09 |
systemdlete | (summer '21) | 20:09 |
fsmithred | yeah, we follow them. They have not even begun freeze on bullseye. | 20:09 |
fsmithred | however, most of the forked packages are already forked and working. | 20:10 |
fsmithred | and the list of forked packages does grow as time goes on. Latest thing is for upstream to drop init scripts from the package. | 20:11 |
systemdlete | So, one could say that Bullseye is a moving target (see what I did there?) | 20:11 |
fsmithred | lol, yeah | 20:11 |
systemdlete | "Latest thing is for upstream to drop init scripts from the package." Does this mean debian has agreed to clean up their packaging to support multiple boot and runtime systems? Or am I reading too much in there? | 20:13 |
fsmithred | I think you're reading it backward. | 20:15 |
systemdlete | So more like "The package drops init scripts for upstream's latest thing?" | 20:16 |
fsmithred | some debian devs are not interested in compatibiity with other init systems | 20:16 |
systemdlete | Actually, fsmithred, it is probably surprising that there is any interest there at all! | 20:16 |
fsmithred | brb need food | 20:17 |
systemdlete | So what did that mean then? | 20:17 |
fsmithred | what? | 20:53 |
mason | systemdlete: Have you followed the recent/relevant Debian bug? https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=975075 | 20:58 |
rwp | systemdlete, Well before the creation of Devuan I decided that pam_systemd.so did things I did not want, and therefore have since removed it from the PAM configuration files. | 21:57 |
rwp | Among other things pam_systemd.so is responsible for creating the /run/user/$(id -u) files. | 21:57 |
rwp | Among other things pam_systemd.so breaks starting X with a permission denied error. X tries to start, clears the screen, then gets stuck. | 21:58 |
rwp | This might be a problem due to the resulting cgroup non-configuration that I imposed by removing it. | 21:58 |
rwp | So I do the Luddite thing and remove it and revert to starting X the old ways. | 21:59 |
rwp | YMMV! | 21:59 |
aitor_ | hi | 23:59 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!