So today I had to make a test change on a brocade HA pair. We updated a match-list not in use and it ended up taking down prod sites despite the change only being a match-list that wasn't assigned to any health check. I verified the changes against the before and after configs. To get traffic back we had to fail over to the other brocade, then restart the first unit.
What could I do to mitigate this in the future? We had to test the production side to make sure we captured the right health check string that's only available in prod so the maintenance prep could be written up. We don't have stage brocades.
Should I:
- Require brocades in HA in stage and make assumptions what prod will check?
- Upgrade to something like F5 to avoid issues with brocades?
- Other?
No comments:
Post a Comment