Bug 50424 - Test setup for domains with mixed UCS releases (updated and new installed)
Test setup for domains with mixed UCS releases (updated and new installed)
Status: NEW
Product: UCS Test
Classification: Unclassified
Component: Jenkins - Test setups
unspecified
Other Linux
: P5 normal (vote)
: ---
Assigned To: UCS maintainers
:
Depends on:
Blocks:
  Show dependency treegraph
 
Reported: 2019-10-28 18:27 CET by Arvid Requate
Modified: 2019-10-28 19:20 CET (History)
2 users (show)

See Also:
What kind of report is it?: Development Internal
What type of bug is this?: ---
Who will be affected by this bug?: ---
How will those affected feel about the bug?: ---
User Pain:
Enterprise Customer affected?:
School Customer affected?:
ISV affected?:
Waiting Support:
Flags outvoted (downgraded) after PO Review:
Ticket number:
Bug group (optional):
Max CVSS v3 score:


Attachments

Note You need to log in before you can comment on or make changes to this bug.
Description Arvid Requate univentionstaff 2019-10-28 18:27:30 CET
We should add a test setup for domains with mixed UCS releases:


UCS DC Master updated e.g. from UCS 4.3-0 to UCS-latest
UCS DC Slave  updated e.g. from UCS 4.3-0 to UCS 4.3-4

and then add a new DC Slave with UCS 4.4-2, updated to latest errata before joining.


We should have this both for unreleased errata and for "only released".


Bug #50342 Comment 3 is an example of a regression that would have been detected in a setup like this.

Setups like this are also common in the lifecycle of UCS domains in customer production environments.
Comment 1 Florian Best univentionstaff 2019-10-28 18:33:14 CET
I don't understand why we should add a UCS 4.4-2 slave and how that would detect the case?
Or do you mean that ucs-test should be executed on every of these hosts? (Currently ucs-test is only executed on one hosts in a ucs-ec2-configuration file).

I think we should join a UCS 4.0-0 Slave into a UCS 4.4-2-latest DC Master and run ucs-test on that DC-Slave.
Comment 2 Arvid Requate univentionstaff 2019-10-28 19:20:03 CET
I already discussed this with Felix:

> I don't understand why we should add a UCS 4.4-2 slave and how that would detect the case?

There are cases, where changes are not activated becasue the joinscript version is not incremented. But when a new slave with 4.4-2 + latest errata joins, then it actually runs the joinscript. In the case of Bug #50342 Comment 3 this would have activated the new version of the udm module and that would make things fail on at least one of the other systems. Felix only deiscovered that regression in a (non-updated) Docker container app, but we don't want to rely on that for the future.


> Or do you mean that ucs-test should be executed on every of these hosts? (Currently ucs-test is only executed on one hosts in a ucs-ec2-configuration file).

Sure. I think there are already test setups where we run at least parts of ucs-test on several machines. But I'm not in implementation mode here, I'm just reporting what needs to be done based on the discussion with Felix.


> I think we should join a UCS 4.0-0 Slave into a UCS 4.4-2-latest DC Master and run ucs-test on that DC-Slave.

As explaind, we thought a while about this and the setup I described is a) realisitc and b) would have detected the Bug class in question. But we can discuss this when implementation starts if you have serious objections. I just needed a bug to not forget about htis and to create a Kanban card.