action #168106
closedcoordination #167054: [epic] Run more workloads in CC-compliant PRG2 to be less affected by CC related network changes
QE PXE server in PRG2
0%
Description
Motivation¶
The non-compliant NUE2 based PXE server qa-jump.qe.nue2.suse.org is relied upon also for PRG2 based machines. In case CC-related network changes due to #165282 might include preventing access from non-CC areas to openqa.suse.de or hosts within CC areas like PowerPC in PRG2 might not be able to access qa-jump.qe.nue2.suse.org anymore which would pose problems with all bare-metal tests, PowerPC and similar. For this setting up a PRG2 local PXE server would be the right approach. This however is already planned with #155524
Acceptance criteria¶
- AC1: PRG2 based openQA tests relying on PXE do not rely on qa-jump.qe.nue2.suse.org
- AC2: NUE2 based openQA tests relying on PXE still use qa-jump.qe.nue2.suse.org
Suggestions¶
- Block on #155524
- Setup a new PXE server based on existing qa-jump using or based on https://gitlab.suse.de/qa-sle/qa-jump-configs/
- Adjust the next_server variable in https://gitlab.suse.de/OPS-Service/salt/ for PRG2 based QE machines accordingly
- Consider the impact of baremetal-support.qe.nue2.suse.org
Updated by okurz about 2 months ago
- Assignee changed from okurz to mkittler
assigning to mkittler to track blocker tickets for now
Updated by szarate about 2 months ago · Edited
For the sake of completeness
https://gitlab.suse.de/OPS-Service/salt/-/merge_requests/5748
If my very naive understanding is correct, all is needed to unblock https://jira.suse.com/browse/ENGINFRA-3941 is the VM + PR above with the proper changes (see the description)
Updated title of https://jira.suse.com/browse/ENGINFRA-3941 so it reflects better what is needed
@Oliver Kurz
can you confirm that it is like that? yes/no?
@Moroni Flores
lmk if there’s an issue with provisioning the VM (id guess same specs/os as qa-jump)
I’d say we can use qamaster if you can’t provision, but I realized it is in NUE, and likely will have the wireguard tunnel… and that defeats the purpose (I guess, but it does not have the tunnel
Updated by okurz about 23 hours ago
- Status changed from Blocked to New
- Priority changed from Normal to High
#155524 was resolved. Please check and ensure that both ACs are covered and if not working please collaborate with dheidler.
Updated by mkittler about 16 hours ago
I'll check documentation mentioned on #155524#note-43 when GitLab comes back and will check what was changed via https://gitlab.suse.de/OPS-Service/salt/-/merge_requests/5893.
It looks like IPXE_HTTPSERVER already points to http://baremetal-support.qe.prg2.suse.org on relevant worker slots (on worker33, 34, 35 and 36). The NUE2-URL http://baremetal-support.qe.nue2.suse.org is only used on sapworker1 slots anymore. So I don't have to update workerconf.sls.
Updated by mkittler about 16 hours ago
- Status changed from New to In Progress
It looks like the PXE setup generally works. There are passing jobs like https://openqa.suse.de/tests/16210082 using the new PXE server according to https://openqa.suse.de/tests/16210082/file/vars.json. Other examples are https://openqa.suse.de/tests/16210189, https://openqa.suse.de/tests/16238719 and https://openqa.suse.de/tests/16239161.
I suppose with that all ACs are fulfilled. AC2 is also fulfilled because the support server in NUE2 is still up and running (and connectivity problems with it are handled in #173839).
Updated by dheidler about 15 hours ago
To my knowledge both ACs are already resolved.