fails to start in unprivileged container
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
corosync (Debian) |
Fix Released
|
Unknown
|
|||
corosync (Ubuntu) |
Fix Released
|
Medium
|
Dan Streetman | ||
Focal |
Fix Released
|
Medium
|
Dan Streetman | ||
Groovy |
Fix Released
|
Medium
|
Dan Streetman | ||
Hirsute |
Fix Released
|
Medium
|
Dan Streetman |
Bug Description
[impact]
fails to start in container
[test case]
install corosync in container and check status
root@corosync-f:~# systemctl status corosync
● corosync.service - Corosync Cluster Engine
Loaded: loaded (/lib/systemd/
Active: failed (Result: exit-code) since Thu 2021-03-11 21:28:52 UTC; 17s ago
Docs: man:corosync
Main PID: 14552 (code=exited, status=15)
Mar 11 21:28:51 corosync-f corosync[14552]: [TOTEM ] Initializing transport (Kronosnet).
Mar 11 21:28:52 corosync-f corosync[14552]: [TOTEM ] knet_handle_new failed: File name too long (36)
Mar 11 21:28:52 corosync-f corosync[14552]: [KNET ] transport: Failed to set socket buffer via force option 33: Operation not permitted
Mar 11 21:28:52 corosync-f corosync[14552]: [KNET ] transport: Unable to set local socketpair receive buffer: File name too long
Mar 11 21:28:52 corosync-f corosync[14552]: [KNET ] handle: Unable to initialize internal hostsockpair: File name too long
Mar 11 21:28:52 corosync-f corosync[14552]: [MAIN ] Can't initialize TOTEM layer
Mar 11 21:28:52 corosync-f corosync[14552]: [MAIN ] Corosync Cluster Engine exiting with status 15 at main.c:1531.
Mar 11 21:28:52 corosync-f systemd[1]: corosync.service: Main process exited, code=exited, status=15/n/a
Mar 11 21:28:52 corosync-f systemd[1]: corosync.service: Failed with result 'exit-code'.
Mar 11 21:28:52 corosync-f systemd[1]: Failed to start Corosync Cluster Engine.
[regression potential]
any regression would likely result in failure to start, or reduced performance of corosync
[scope]
this is needed in f/g/h
corosync starts in a bionic container
opened upstream PR
https:/
this is also related to bug 1911904 and bug 1828228
Related branches
- Bryce Harrington (community): Approve
- git-ubuntu developers: Pending requested
-
Diff: 368 lines (+240/-9)7 files modifieddebian/changelog (+197/-0)
debian/control (+5/-2)
debian/patches/Make-the-example-config-valid.patch (+13/-5)
debian/patches/lp1918735/0001-allow_knet_handle_fallback_default_yes.patch (+22/-0)
debian/patches/series (+1/-0)
debian/tests/control (+1/-1)
debian/tests/quorumtool (+1/-1)
description: | updated |
description: | updated |
description: | updated |
Changed in corosync (Debian): | |
status: | Unknown → New |
Changed in corosync (Debian): | |
status: | New → Fix Released |
I can confirm the issue comparing LXD to VM guest.
But given that the errors are about corosync being unable to set some device options - and not having an obvious "container mode" switch I wonder if that will end up needing upstream work.
Also FYI this seems to be a bit of a re-surface of bug 1828228 /bugs.launchpad .net/auto- package- testing/ +bug/1828228/ comments/ 8
=> https:/
@Dan - are you intending to work-on/fix this with your SEG hat on or did you just spot and wanted to report it for us?