+++ This bug was initially created as a clone of Bug #1897543 +++
+++ This bug was initially created as a clone of Bug #1897542 +++
+++ This bug was initially created as a clone of Bug #1895509 +++
Description of problem:
A recent change in 4.5 introduced a bug that disallows backups taken on one master to be restored on other masters.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. Take a cluster-backup using cluster-backup.sh on one master.
2. Copy the backup on to another master.
3. Attempt to restore the database on the other master using procedure documented.
The etcds fail to come up on all masters.
All etcds successfully start on all masters and the cluster recovers.
--- Additional comment from Suresh Kolichala on 2020-11-06 20:40:14 UTC ---
As a workaround, a backup should be restored on the same master it is taken from.
To determine the master where the backup is taken from, one may run the following command against the backup directory:
sudo tar xvzf <backup>/static_kuberesources_*.tar.gz *restore-etcd-pod/pod.yaml --to-stdout 2>&1 | grep 'ETCD_NAME='
Tried with 4.4.0-0.ci.test-2020-12-23-042527-ci-ln-cs764w2
got backup from master-0, then restore with this backup on master-2, it succeed.
# oc get pods -n openshift-etcd | grep etcd
etcd-ip-10-0-139-137.us-east-2.compute.internal 4/4 Running 0 3m34s
etcd-ip-10-0-168-222.us-east-2.compute.internal 4/4 Running 0 2m33s
etcd-ip-10-0-205-133.us-east-2.compute.internal 4/4 Running 0 3m3s
Verified at comments 3, but I just curious about why bug status have not updated to 'verified' since I have executed pre-merge verification.
Since the problem described in this bug report should be
resolved in a recent advisory, it has been closed with a
resolution of ERRATA.
For information on the advisory (Important: OpenShift Container Platform 4.4.33 bug fix and security update), and where to find the updated
files, follow the link below.
If the solution does not work for you, open a new bug report.