search

SUSE / velum

Dashboard for CaaS Platform clusters (v1, v2 and v3)
https://www.suse.com/
Apache License 2.0
54 stars 30 forks source link

[3.0] Fix pillar issue in public cloud #628

Closed jgleissner closed 6 years ago

jgleissner commented 6 years ago

I've added a call_runner function to salt interface and added a call to sync_pillar runner in cloud_cluster.rb. This is required to activate the external pillar, which is normally done at orchestration time, but for creating cluster nodes in the cloud, the pillar is needed before that. See also https://github.com/kubic-project/velum/issues/399 for details.

I've also added a sleep(1) after the refresh_pillar, as we ran into the issue that the succeeding salt-cloud still works on the old pillar data, and this seems to work around it. I know that this is a bit ugly and perhaps can be fixed better in salt directly, but I haven't had any luck yet with narrowing the issue down.

kiall commented 6 years ago

Hey @jgleissner - is there a matching change to master? I don't see one pending / merged.

jgleissner commented 6 years ago

Hi @kiall, I've opened a PR against master: https://github.com/kubic-project/velum/pull/636

jordimassaguerpla commented 6 years ago

@jgleissner : The PR for master was merged. Please backport your fixes to this PR. Thanks.

https://github.com/kubic-project/velum/pull/636

jgleissner commented 6 years ago

Hi @jordimassaguerpla , I've cherry-picked the remaining commits from the master PR.

jordimassaguerpla commented 6 years ago

I saw some errors when testing the cluster in jenkins. I just retriggered the tests to make sure these are not random.