Package Details: nbdkit 1.40.4-1

Git Clone URL: https://aur.archlinux.org/nbdkit.git (read-only, click to copy)
Package Base: nbdkit
Description: NBD server toolkit
Upstream URL: https://gitlab.com/nbdkit/nbdkit/
Licenses: custom: BSD
Submitter: t-8ch
Maintainer: t-8ch (rwmjones)
Last Packager: t-8ch
Votes: 5
Popularity: 0.000419
First Submitted: 2019-02-22 14:36 (UTC)
Last Updated: 2024-10-13 10:57 (UTC)

Latest Comments

1 2 3 4 5 6 .. 12 Next › Last »

aaruni96 commented on 2024-10-17 10:42 (UTC)

No, I don't think I have SELinux/Apparmor/seccomp enabled at all. (Its vanilla arch, not manjaro or endeavor or whatever, and I assume those things are not enabled if I didn't explicitly set those up.)

rwmjones commented on 2024-10-17 10:40 (UTC)

Could it be a SELinux / Apparmor / seccomp type of thing going on? That's the only thing I can think of that would stop nbdkit from opening an unused port.

Anyway tests that use pick_unused_port can all be skipped safely.

aaruni96 commented on 2024-10-17 10:30 (UTC)

You are right @rwmjones, the update didn't help. But also, I don't understand why the port picking is failing.

Here are my open ports :

% sudo lsof -nP -iTCP -sTCP:LISTEN
COMMAND    PID            USER  FD   TYPE DEVICE SIZE/OFF NODE NAME
systemd-r  854 systemd-resolve  12u  IPv4  21798      0t0  TCP *:5355 (LISTEN)
systemd-r  854 systemd-resolve  14u  IPv6  21806      0t0  TCP *:5355 (LISTEN)
systemd-r  854 systemd-resolve  21u  IPv4  21813      0t0  TCP 127.0.0.53:53 (LISTEN)
systemd-r  854 systemd-resolve  23u  IPv4  21815      0t0  TCP 127.0.0.54:53 (LISTEN)
sshd       930            root   7u  IPv4  24125      0t0  TCP *:22 (LISTEN)
sshd       930            root   8u  IPv6  24127      0t0  TCP *:22 (LISTEN)
tailscale  932            root  20u  IPv6  10426      0t0  TCP [fd7a:115c:a1e0::1]:59780 (LISTEN)
tailscale  932            root  21u  IPv4  10427      0t0  TCP 100.64.0.1:58436 (LISTEN)
kdeconnec 1490          aaruni 120u  IPv6   3750      0t0  TCP *:1716 (LISTEN)
kdeconnec 1490          aaruni 128u  IPv6  59807      0t0  TCP *:1745 (LISTEN)
kdeconnec 1490          aaruni 142u  IPv6  89356      0t0  TCP *:1739 (LISTEN)
kdeconnec 1490          aaruni 143u  IPv6  83491      0t0  TCP *:1740 (LISTEN)
kdeconnec 1490          aaruni 144u  IPv6  83494      0t0  TCP *:1741 (LISTEN)
kdeconnec 1490          aaruni 150u  IPv6  57751      0t0  TCP *:1742 (LISTEN)
kdeconnec 1490          aaruni 151u  IPv6  25494      0t0  TCP *:1743 (LISTEN)
kdeconnec 1490          aaruni 152u  IPv6  25500      0t0  TCP *:1744 (LISTEN)
kdeconnec 1490          aaruni 154u  IPv6  88867      0t0  TCP *:1746 (LISTEN)
kdeconnec 1490          aaruni 155u  IPv6 106842      0t0  TCP *:1747 (LISTEN)
kdeconnec 1490          aaruni 156u  IPv6  94710      0t0  TCP *:1748 (LISTEN)
kdeconnec 1490          aaruni 157u  IPv6  98850      0t0  TCP *:1749 (LISTEN)

Here's the port the test tried to use :

+ pick_unused_port
+ requires ss --version
+ port=62877
+ ss -ltn
+ grep -sqE ':62877\b'
+ echo picked unused port 62877
picked unused port 62877
++ nbdkit -i localhost -p 62877 -e 123456781234567<very long string> null --run 'echo "$uri"'
bind: Cannot assign requested address

Looks like port 62877 chosen by the test is actually unused ?

I can disable the test, or we can keep using my machine to try to debug this, whichever you prefer.

rwmjones commented on 2024-10-13 12:21 (UTC) (edited on 2024-10-13 12:22 (UTC) by rwmjones)

I don't think updating will help. What's happening is that we need to test using a TCP port, and all we can do is pick one at random and hope for the best. I suppose in this case it's either not choosing the port randomly for some reason (some problem in bash?) or all your ports are already used. Anyway you can just skip the test by doing:

rm tests/test-long-name.sh
touch tests/test-long-name.sh
chmod +x tests/test-long-name.sh
make check

t-8ch commented on 2024-10-13 10:57 (UTC)

@aaurni96 The version is now up to date.

aaruni96 commented on 2024-10-13 08:49 (UTC)

Yes, it fails at that exact spot every time over multiple tries.

Could it have something to do with the AUR version not being up to date with upstream (1.40.2-1 v/s 1.40.4) ?

rwmjones commented on 2024-10-10 21:52 (UTC)

Is the failure repeatable? Unfortunately a few tests including this one have to randomly assign a public port using this function, and it's not free of races: https://gitlab.com/nbdkit/nbdkit/-/blob/3cc7389e60da67fd29ed708c987d686227e3abf2/tests/functions.sh.in#L527

aaruni96 commented on 2024-10-10 21:28 (UTC)

Nbdkit fails to pass test-long-name.sh for me.

Error for nbdkit -i localhost -p 55951 -e <really-long-label> is

bind: Cannot assign requested address

rwmjones commented on 2024-05-28 07:50 (UTC)

Re: test-gcs:

We recently attempted to fix this test, see: https://gitlab.com/nbdkit/nbdkit/-/commit/aca30111632091bbc91096f19638b9e1df7f02b6. However that commit is included already in 1.38.2, so I don't know what's going on there. You might just try skipping that test.

Re: cargo --offline:

I've no idea about this, suggestions welcome. For Fedora we disable rust bindings completely. Rust & Golang have such weird language packaging that we feel it's best if developers who care about it build it for themselves.

solsticedhiver commented on 2024-05-28 06:59 (UTC)

I built it in a clean chroot like said in https://wiki.archlinux.org/title/DeveloperWiki:Building_in_a_clean_chroot and this works

it failed outside of it, I think, because I have python-protobuf installed too. I can run import googl in a python session, and this works. No init.py or whatsoever because I remove src direcotry without build.