#archlinux32 | Logs for 2023-02-09
Back
[00:29:17] -!- morriset has quit [Quit: Leaving]
[03:00:15] -!- lithiumpt has quit [Ping timeout: 248 seconds]
[03:11:11] -!- bill-auger_ has joined #archlinux32
[03:11:25] -!- bill-auger_ has quit [Client Quit]
[03:11:44] -!- bill-auger_ has joined #archlinux32
[03:12:02] -!- bill-auger_ has quit [Client Quit]
[03:13:01] -!- bill-auger_ has joined #archlinux32
[03:15:50] -!- bill-auger has quit [*.net *.split]
[03:15:50] -!- abouvier has quit [*.net *.split]
[03:18:49] -!- bill-auger_ has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
[03:19:16] -!- bill-auger has joined #archlinux32
[03:27:11] -!- abouvier has joined #archlinux32
[03:30:24] -!- drathir_tor has quit [Remote host closed the connection]
[03:35:48] -!- drathir_tor has joined #archlinux32
[05:59:10] -!- lithiumpt has joined #archlinux32
[06:10:50] -!- GNUtoo has quit [Ping timeout: 255 seconds]
[06:16:26] -!- GNUtoo has joined #archlinux32
[07:23:31] -!- titus_livius has joined #archlinux32
[07:23:47] -!- buildmaster has quit [Read error: Connection reset by peer]
[07:26:29] -!- buildmaster has joined #archlinux32
[07:44:26] -!- jonathon has joined #archlinux32
[07:44:29] -!- jonathon has quit [K-Lined]
[08:19:48] -!- City-busz[m] has quit [Quit: Client limit exceeded: 20000]
[11:33:29] -!- drathir_tor has quit [Ping timeout: 255 seconds]
[11:38:11] -!- epony has quit [Remote host closed the connection]
[11:38:40] -!- epony has joined #archlinux32
[11:40:27] -!- City-busz[m] has joined #archlinux32
[11:42:44] -!- drathir_tor has joined #archlinux32
[12:14:02] -!- City-busz[m] has quit [Ping timeout: 265 seconds]
[12:29:09] -!- City-busz[m] has joined #archlinux32
[14:09:38] -!- drathir_tor has quit [Ping timeout: 255 seconds]
[14:11:45] -!- drathir_tor has joined #archlinux32
[16:32:17] <girls> gpg: trustdb rec 70: write failed (n=-1): Bad file descriptor
[16:32:17] <girls> gpg: Fatal: /etc/pacman.d/gnupg/trustdb.gpg: failed to append a record: Bad file descriptor
[16:32:22] <girls> ^ on the buildmaster
[16:32:28] <girls> are the disks dying?
[16:32:52] <girls> side note: I recreated /etc/pacman.d/gnupg before, because I saw this error
[16:33:02] <girls> ... error still present :-/
[16:57:43] <bill-auger> i get that error every few months or so
[16:58:54] <bill-auger> i dont know why it happens but it is always in the same circumstance: checking a signature as a non-root user
[16:59:12] <bill-auger> i discovered that it is fixed easily by checking any signature as root - that causes the tristdb to be updated
[16:59:32] <bill-auger> checking a signature with `pacman-key --verify`
[17:01:51] <bill-auger> the error message is very misleading - there is nothing wrong with the disk - its a permissions problem
[17:03:02] <bill-auger> but it maybe is not really a problem - non-root user should not be able to modify the trustdb
[17:03:50] <bill-auger> when the trustdb is updated, you will see a message like "next trustdb check due on <some_date>" - i suspect this is actually expected behavior
[19:35:58] -!- drathir_tor has quit [Remote host closed the connection]
[19:56:27] -!- drathir_tor has joined #archlinux32
[20:25:25] <girls> Thanks, bill-auger - that really is the issue.
[20:26:10] <girls> "next trustdb check due at 2023-12-31" - we will se us again at new years eve
[20:28:40] -!- pants has quit [Quit: bye]
[20:34:58] <T`aZ> just to be pedantic, would not this be a tiny bug in gpg, not reporting the correct error message ?
[20:36:07] -!- pants has joined #archlinux32
[20:36:30] <girls> it reports a warning (when it's working), though :D
[20:38:26] <T`aZ> but it should say 'permission denied' or related, no ?
[20:40:11] <girls> at least it doesn't land in the logs of the scripts - maybe some stdout vs. stderr thingy (but I thought, we captured both)
[20:56:11] <bill-auger> (donning my nerd hat) i think the deal may be that GPG runs in a very isolated memory space without full acces to the filesystem - it gets the FD "passed in" somehow, so the FD is null if the user can not "see" the file before the command executes
[20:56:38] <bill-auger> is that theory too far-fetched maybe?
[20:57:21] <bill-auger> o/c still the error message could be better
[20:57:57] <bill-auger> but it would not know the difference between "file not found" and "permission denied"
[21:02:12] <bill-auger> the best error message may be "i cant let you do that, dave"
[21:11:21] <bill-auger> girls: youre probably not capturing stderr - i see it in the systemd journal - thats how i discovered it, and guessed that i should try to run the same --verify command as root, to see if root got the same error
[21:12:29] <girls> find ... -exec gnupg --verify '{}.sig' '{}' \; 2> /dev/null
[21:12:37] <girls> looks, like we should capture it
[21:12:52] <girls> though, I'm not certain, what "find" does with stderr in that case ...
[21:13:14] <bill-auger> when root not not get the same error, it still was not obvious that it fixed the problem - i only noticed the next day that it started working again magically
[21:13:29] <girls> :D
[21:20:01] <bill-auger> ok i had to look - this is where i see it (in dbscripts)
[21:20:01] <bill-auger> if ! pacman-key -v "${pkg}.sig" >/dev/null 2>&1; then
[21:20:01] <bill-auger> die "Package %s does not have a valid signature" "$repo/${pkg##*/}"
[21:21:27] <bill-auger> so you see, the failure was not evident t me either - i added some code before the `die` command to run the --verfy command again, so it does go into the logs now - i just did not remember
[21:25:42] <bill-auger> i did not remember writing that extra code - and ive still never published it to VCS - i should do that - it is just as poor of an error message; because it is not a bad signature
[21:26:42] <bill-auger> o/c it _could_be_ a bad signature - it just usually is not
[21:29:14] <bill-auger> it arguably a worse error message, because it could be correct, but usually is not
[22:12:18] <bill-auger> coincidence of coincidences - i just hit that error
[23:39:55] -!- epony has quit [Quit: QUIT]