Bug 7857 - Subcluster info can be out of sync between tlwebadm and vsmserver
Summary: Subcluster info can be out of sync between tlwebadm and vsmserver
Status: CLOSED FIXED
Alias: None
Product: ThinLinc
Classification: Unclassified
Component: Other (show other bugs)
Version: trunk
Hardware: PC Unknown
: P2 Normal
Target Milestone: 4.15.0
Assignee: William Sjöblom
URL:
Keywords: ossman_tester, prosaic
Depends on:
Blocks: 7589
  Show dependency treegraph
 
Reported: 2022-03-09 15:48 CET by Linn
Modified: 2023-08-30 14:27 CEST (History)
2 users (show)

See Also:
Acceptance Criteria:


Attachments

Description Linn cendio 2022-03-09 15:48:15 CET
Currently, we only track which agents that belong to a certain subcluster in vsmserver.hconf. 

This can lead to inconsistencies between vsmserver and tlwebadm/tlctl if the conf have been updated but the vsmserver service not yet has been restarted. Ideally, we would like to store this information in a way where this inconsistency can't happen.
Comment 1 Pierre Ossman cendio 2022-04-14 08:34:30 CEST

*** This bug has been marked as a duplicate of bug 7589 ***
Comment 2 William Sjöblom cendio 2022-04-14 10:21:49 CEST
This is not a duplicate of bug 7589 since this bug only concerns tlwebadm while bug 7589 also concerns `tl-ssh-all' and `tl-rsync-all'.

The scope of this bug was fixed in bug 7589, so let's close this bug as fixed (as a blocker of that bug) and let that bug remain open.
Comment 3 William Sjöblom cendio 2022-04-14 10:26:12 CEST
Apparently this bug was untested, so let's mark it as resolved instead.
Comment 4 William Sjöblom cendio 2022-04-14 10:30:23 CEST
To clarify, the changes to be tested here are those concerning tlwebadm in bug 7589 (i.e. all changes made for that bug as of writing this).
Comment 5 Pierre Ossman cendio 2022-04-14 11:22:45 CEST
Tested build 2582 on RHEL 7. I can see that it respects what vsmserver sees, not the config:

1. Configured two subclusters, restarted everything
2. Removed one subcluster, tlwebadm still shows two
3. Restarted just vsmserver, tlwebadm now shows one subcluster
4. Same thing, but adding the missing subcluster again

All load information is also still shown correctly, so no regression there. “tlctl load” also still works.

Error handling code was also touched, so I tested stopping vsmserver and tlwebadm still correctly shows that it gets a connection error.

Note You need to log in before you can comment on or make changes to this bug.