-
Bug
-
Resolution: Done
-
Undefined
-
4.12.0
Description of problem:
While installing cluster with assisted installer lately we have cases when one of the master joins very quickly and start all needed pods in order for cluster bootstrap to finish but the second one joins only after that. Keepalived can't start if there is only one joined cluster as it doesn't have enough data to build configuration files. In HA mode cluster bootstrap should wait at least for 2 joined masters before removing bootstrap control plane as without it installation with fail.
Version-Release number of selected component (if applicable):
How reproducible:
Start bm installation and start one master, wait till it starts all required pods and then add others.
Steps to Reproduce:
1. Start bm installation 2. Start one master 3. Wait till it starts all required pods. 4. Add others
Actual results:
no vip, installation fails
Expected results:
installation succeeds, vip moves to master
Additional info:
- blocks
-
OCPBUGS-10923 Cluster bootstrap waits for only one master to join before finishing
- Closed
- is cloned by
-
OCPBUGS-10923 Cluster bootstrap waits for only one master to join before finishing
- Closed
- is related to
-
OCPBUGS-2852 etcd failure: failed to make etcd client for endpoints [https://[2620:52:0:1eb:367x:5axx:xxx:xxx]:2379]: context deadline exceeded
- Closed
- links to