cores | good evening. i installed devuan onto a virtualbox vm and after setting up i did "sudo apt install fluxbox" but apt is stuck on "Setting up xfonts-terminus" for about 10 minutes. | 04:41 |
---|---|---|
cores | is something broken or should i just kill it and retry? any ideas on how i can find the problem? | 04:41 |
cores | daedalus 5.0 | 04:41 |
cores | there doesn't appear to be a lot of network or cpu activity | 04:43 |
rwp | If it were me I would log in with another terminal and then "ps -efH | less" and look to see what the process tree looks like around apt. | 04:48 |
rwp | If it looks like it really is hung and not doing anything then yes I would kill it and restart it. The apt state might be unfinished but it will resume and continue. | 04:49 |
cores | good idea. apt has a few child processes still going: '/usr/bin/dpkg --status-fd 16 --configure --pending' and '/bin/sh /var/lib/dpkg/info/xfonts-terminus.postinst configure' and 'xset fp rehash' | 04:52 |
rwp | That makes it appear that it is waiting for input. Maybe? | 04:52 |
cores | those are a tree of processes, not all spawned by apt | 04:53 |
cores | waiting for input? hmm... | 04:53 |
cores | it could be, but i don't see any place to put input | 04:54 |
rwp | The file /var/lib/dpkg/info/xfonts-terminus.postinst is simply a script that you can browse. I often look. Sometimes it is obvious what the postinst are doing. Sometimes someone has programmed a nightmare into it and it is not possible to easily discern this. | 04:54 |
rwp | If interrupted then dpkg will record the state as failed and needs-configure. Then any dpkg operation will trigger it to "dpkg --configure -a" and that will reconfigure all unconfigured packages. | 04:55 |
rwp | Or one can "apt-get install -f" to run the dpkg --configure and also possibly try to fill any missing dependencies. Same thing really but I do the dpkg --configure -a if I am resuming so I can see any messages dpkg is putting out in isolation. | 04:56 |
rwp | In this case it isn't fluxbox itself that is having a problem but xfonts-terminus, a font that I also have installed, no problem here. | 04:57 |
rwp | Maybe the installation at that point became resource starved for memory? In a VM without a lot of memory I could see that being a problem. Is there swap configured? Often configuring just a little swap can give just enough headroom to make all of the difference. | 04:58 |
cores | weird, i did "view /var/lib/dpkg/info/xfongs-terminus.postinst" and that command didn't do anything | 04:58 |
rwp | s/fongs/fonts/ :-) | 04:58 |
cores | right | 04:58 |
cores | i can't copy/paste yet | 04:58 |
cores | the 'view' command is just sitting there, it's not showing me the file | 04:59 |
cores | in terms of memory i'm at 1.3 GB used of 2 GB | 04:59 |
cores | it's not a lot of memory, i know | 04:59 |
cores | I should bump that | 04:59 |
rwp | It's crazy that 2GB might not be sufficient. But I would suggest adding some swap to supplement. | 05:00 |
cores | so 'view' didn't work, but 'vi' did | 05:00 |
rwp | I looked at that postinst here and literally it has two commands in it. "xset fp rehash" which I did not think would work without DISPLAY set so that's odd. | 05:01 |
rwp | And "update-fonts-dir --x11r7-layout misc;update-fonts-alias --include /etc/X11/fonts/misc/xfonts-terminus.alias misc" which is actually doing the work. | 05:01 |
cores | i have the same thing | 05:01 |
rwp | When you get the spare moment you might run "update-alternatives --display view" to look to see what is providing the view command. | 05:01 |
cores | it appears "xset fp rehash" is the thing that stuck | 05:01 |
rwp | Also let me tease you about using view when "less" is more than "more". :-) (Hey, I am just joking because I emacs'd that file here to look.) | 05:03 |
cores | "update-alternatives --display view returns a bunch of stuff, it starts with "view - auto mode" and "link best version is /usr/bin/mcview" which is a link to /usr/bin/mc | 05:04 |
rwp | That postinst script has <<if [ "$1" = "configure" -a -n "$DISPLAY" ]; then>> so it would only run that if DISPLAY is set. And won't run it if it is not set. I would unset DISPLAY! At least for the run. It does not need DISPLAY set to install the fonts. That's just silly. | 05:04 |
rwp | On your system, and mine too as I look, view is set to midnight-commander. | 05:04 |
rwp | But mc should still work just fine. It's a text-mode very mature program. | 05:05 |
rwp | TIL that midnight-commander has a vi emulation mode. I did not expect that one. | 05:05 |
cores | ok so kill the apt process, unset DISPLAY (even though I'm logged into X) then run the postinst script? | 05:05 |
rwp | I would Control-C out of the install process. Then simplest is: env -u DISPLAY dpkg --configure -a | 05:06 |
rwp | After that completes, and I think it should complete okay, then repeat your previous apt-get install fluxbox command and it should finish doing everything. | 05:07 |
cores | wow, Ctrl-C isn't working on this install process | 05:07 |
cores | I guess I have to kill the terminal window | 05:07 |
rwp | You don't have any keyboard input into it? It isn't responding? You can find the process id in the other window and kill it by pid. | 05:08 |
rwp | I would be worried that if you killed the window that it still would not be dead. Better to kill it by pid and then it will return to the command line and you will see it. Than to have to ps -ef|grep for it. | 05:09 |
cores | i killed it using kill | 05:11 |
rwp | I think that was your only reasonable choice. I don't know why it would have hung. I would review /var/log/syslog and see if any clues were logged there. | 05:12 |
rwp | There is also /var/log/dpkg.log and /var/log/apt/term.log too but I doubt there will be anything of note there beyond what you already know. | 05:13 |
cores | what paste service should i use? | 05:14 |
rwp | If paste.debian.net is resolving for you then that one is good. (It was having a dns problem earlier this weekend.) | 05:14 |
cores | ok starting firefox maybe blew my memory | 05:14 |
cores | basically i was going to paste the output from "sudo env -u DISPLAY dpkg --config -a" | 05:15 |
rwp | A command line way to post things to a pastebin is: echo foo foo foo | nc termbin.com 9999 | 05:15 |
cores | "dpkg: error: dpkg database was locked by another process..." | 05:15 |
rwp | Waiting for a semaphore. That would do it. But why? Some other process running. Odd. | 05:15 |
rwp | I would think based upon that information that you had run two commands to do something overlapping and one was waiting for input and the other was waiting for the first. | 05:16 |
rwp | No matter now. Kill all of them. Get to a state where none of them are running. Then run the sequence I listed above. All will be okay. | 05:16 |
cores | https://termbin.com/cnw4 | 05:17 |
cores | Ok, let me try | 05:17 |
rwp | Oh good. I see that https://dnsviz.net/d/snow-crash.org/dnssec/ (hosting paste.debian.net) fixed the DNS issues it was having this weekend. (You can see the errors from this weekend at https://dnsviz.net/d/snow-crash.org/ZdKH7A/dnssec/) | 05:18 |
cores | https://termbin.com/jt33 | 05:19 |
cores | maybe that explains things? two commands for the same thing? | 05:19 |
rwp | Since it says the other pid is 4575 you can look for it "ps -efH | less +/4575" | 05:19 |
cores | maybe i did it by accident | 05:19 |
cores | killed 4575 | 05:20 |
cores | ran "sudo env -u DISPLAY dpkg --config -a" and it's finished | 05:20 |
rwp | \o/ Almost done then. | 05:20 |
cores | the apt progress bar is still at the bottom of the terminal, last line. lol | 05:20 |
rwp | Repeat your original command to install fluxbox and let it finish. | 05:20 |
cores | https://termbin.com/trfl | 05:22 |
cores | i should have just irc'd in from the vm. that would have been easier | 05:22 |
rwp | Looks good to me. Give it a go! | 05:22 |
rwp | Somewhere along the way you ended up with multiple dpkg programs running. dpkg is the underlying engine that is handling package management under apt (and apt-get and aptitude and ...) and that caused it to block waiting for the semaphore as it is supposed to do. Since you had lost connection to that install process killing it by pid made the most sense to clear all of those out of the way. | 05:24 |
cores | i'm in fluxbox now :) | 05:24 |
rwp | apt/dpkg is quite smart enough to understand when it gets interrupted and will have recorded the state correctly allowing it to continue on the next invocation and finish things off. | 05:24 |
rwp | Yay! Good deal. Glad to hear you have everything going. :-) | 05:25 |
cores | rwp, you're a genius. thanks. | 05:25 |
rwp | There's a fine line between genius and insanity. I have erased this line. :-) | 05:26 |
cores | oh one more thing, after i installed devuan i had to manually removme the cdrom from apt sources and add the deb.devuan.org | 05:27 |
cores | not a big deal, just didn't expect it since the rest of the install process went so smooth | 05:27 |
cores | oh and firefox never launched, so i guess i need to bump the RAM | 05:27 |
rwp | That has always been an odd thing that the debian-installer leaves the cdrom line in the sources.list file. I have never understood that either. And I always delete that line too. Cleaning up the sources.list to just what I need. And usually adding non-free contrib as well. | 05:28 |
rwp | If you have the disk space then I would recommend adding at least 1GB of swap. That does not mean it will be used. But it will be counted in the available virtual memory and allow the kernel to avoid resource starvation in many cases that it would fail completely otherwise. | 05:29 |
cores | yeah, i do have 1GB of swap | 05:29 |
rwp | Look for OOM out-of-memory killer incidents logged to the /var/log/syslog file. | 05:29 |
cores | next time I will, I just got impatient and logged out of Gnome | 05:30 |
rwp | Good. People often have the mistaken idea that if they configure swap that the kernel is immediately going to start reading and writing it endlessly just because it is there. And that's not how it works at all. | 05:30 |
cores | what about adding random ubuntu launchpad ppas into my apt sources on devuan? is that a good idea? | 05:30 |
rwp | My advice is: https://wiki.debian.org/DontBreakDebian#Don.27t_make_a_FrankenDebian | 05:31 |
rwp | Ubuntu would be worse than mixing release suites in Debian. Debian and Devuan *are* the same. For all but about a hundred packages forked due to systemd all of the rest of the packages are verbatim Debian so no problem there. But Ubuntu will depend upon a different libc just for the start of things. And if not libc then one of the godzillian other libraries. | 05:32 |
cores | Sounds like good advice | 05:33 |
cores | My host environment is an Ubuntu laptop. I'm running devuan on a VM to see if it can do what I need. | 05:33 |
rwp | If there is something you want from Ubuntu I would ask here first because almost certainly someone will know what is the better answer. If I were going to install from Ubuntu I would do so in a chroot container, but perhaps that is an advanced thing. Compiling from source is not that hard and that would always be better too. Do your own backport of the package. | 05:33 |
cores | My Ubuntu system has all sorts in the apt sources, it's a mile long | 05:33 |
cores | But I don't need all that stuff anyway. I need to cut down. | 05:34 |
rwp | Ubuntu's sources.list file has always been ten miles long. It's a rather hectic collection of this and that which they assemble together. | 05:34 |
rwp | Meanwhile my Devuan systems have three "deb" lines and the same number of matching deb-src lines. | 05:35 |
cores | i can use chroot and/or build from sources | 05:36 |
cores | i prefer not to usually | 05:36 |
cores | i just want to enter one command and go | 05:36 |
cores | that's why i love my mac. as long as i don't try to get too clever, everything is super slick | 05:36 |
rwp | For the most part Ubuntu and Debian/Devuan will have a similar number of packages precompiled and available for immediate installation just the same as each other. | 05:37 |
cores | right | 05:37 |
cores | that's what I expect | 05:37 |
cores | i think that's enough fun for one evening | 05:37 |
cores | thanks for your help | 05:37 |
rwp | Sure thing. Good luck! Enjoy! | 05:38 |
spine-o-saurus | omg it's up and connecting now | 09:16 |
nemo | so... we came up with another solution, but if it's ever a problem in the future | 19:42 |
nemo | what's the correct setup for accessing oldstable in devuan? | 19:42 |
nemo | is there a dedicated archive server for it like there is in debian? | 19:43 |
nemo | or is it on the main apt server url. or should I just use debian's archive server. | 19:43 |
gnarface | afaik, oldstable and oldoldstable are both still in the main repos | 19:43 |
nemo | gnarface: kk thanks. I guess that makes the redirector easier for y'all | 19:44 |
gnarface | there is an archives server url too, jessie is in there and maybe ascii too? | 19:44 |
gnarface | i forget the exact host name | 19:44 |
gnarface | maybe just archive.devuan.org or something like that | 19:44 |
gnarface | but yea, there's a devuan archive server, but oldstable is not old enough for it yet | 19:45 |
nemo | oh. I had to manually copy in tomcat init scripts from other machines again | 19:54 |
nemo | what was that process to see if the scripts were in that devuan package for maintaining scripts? | 19:54 |
nemo | I mean, copying it was not a big deal but I'd rather do things correctly | 19:54 |
gnarface | i forget, there was a package somewhere.... | 20:00 |
u-amarsh04 | gnarface orphan-init-scripts? | 20:14 |
u-amarsh04 | package orphan-sysvinit-scripts | 20:14 |
gnarface | something like that | 20:15 |
gnarface | was it in proposed-updates or something too? | 20:15 |
gnarface | oh, no i see it in main actually | 20:15 |
gnarface | orphan-sysvinit-scripts - Orphaned System-V-like init scripts | 20:15 |
gnarface | at least as of daedalus anyway | 20:15 |
nemo | gnarface: so if I wanted to see if tomcat was in there now, I'd just try reinstalling it? | 20:22 |
nemo | hm. I guess I should move my manually added one out of the way before testing | 20:22 |
gnarface | nemo: no reason you'd have to actually install it. you can tell apt to just download and unpack it into a directory | 20:24 |
gnarface | nemo: or you could use apt-file to search it remotely | 20:24 |
gnarface | that said, it may be that it just puts the missing scripts in /usr/share/doc/, i don't actually know | 20:30 |
nemo | gnarface: welp. tomcat9 is in the changelog at least, but not tomcat10 | 21:20 |
nemo | https://metadata.ftp-master.debian.org/changelogs//main/o/orphan-sysvinit-scripts/orphan-sysvinit-scripts_0.14_changelog | 21:20 |
nemo | I'd swear 9 didn't add an init though | 21:20 |
nemo | (I was installing 9 and 10) | 21:21 |
nemo | yeah. no 10. boo. | 21:22 |
nemo | wow. 9 was added 3 years ago | 21:22 |
gnarface | nemo: didn't it turn out they were basically the same though? | 21:28 |
nemo | gnarface: yeeeah. I just was hoping there was something more automated. I like having distros manage these things, and also it's awkward for the other half dozen coworkers I kinda snuck onto devuan by offering to let them clone my machine | 21:46 |
nemo | ok, step 1, apt install tomcat9 step 2.... | 21:46 |
nemo | but. yes. I just did a diff and the one I "made" that seems to be working is simply a %s/tomcat9/tomcat10/g | 21:47 |
gnarface | nemo: yea, i understand. you might want to try bringing it up with fsmithred again, though i can also understand the apparent absurdity from a package manager's perspective of keeping two separate init scripts just for that one difference | 22:03 |
gnarface | (it's foggy to me, but i seem to recall that being the stated justification for not including both last time, plus something else about the project's own docs maybe insisting they're identical despite the obvious actual difference?) | 22:04 |
gnarface | i dunno for sure, my imagination might be filling in empty blanks with nonsense | 22:05 |
Generated by irclog2html.py 2.17.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!