Add redis.staticID feature to allow constant sentinel ids by manisha-tanwar · Pull Request #191 · StackStorm/stackstorm-k8s

@manisha-tanwar

Redis sentinel not able to elect master & all pods goes into CrashLoopBackOff when worker nodes are rebooted or 2 pods crashes for any reason.
helm/charts#19059

@pull-request-size bot added the size/XS

PR that changes 0-9 lines. Quick fix/merge.

label

Apr 29, 2021
After 2 redis pods failure or if worker nodes are restarted, sentinel not able to elect master.

This feature was added with PR: helm/charts#19059

Sentinels never forget about replicas of a given master, even when they are unreachable for a long time.
In Kubernetes, unless specified, each pod is given a random IP. In our case, with 2 slaves and one master, each replicas is attributed to a random IP. Holding on to old replicas IPs will only make it impossible for other sentinels to elect a new master. constant sentinel ID will allow other sentinel peers to update to the new given IP.

@manisha-tanwar

@manisha-tanwar

manisha-tanwar

@manisha-tanwar manisha-tanwar changed the title Update redis sentinel.staticID: true Add redis.staticID feature to allow constant sentinel ids

Apr 29, 2021

@manisha-tanwar

Move new change to "In Development"

@pull-request-size pull-request-size bot added size/S

PR that changes 10-29 lines. Very easy to review.

and removed size/XS

PR that changes 0-9 lines. Quick fix/merge.

labels

Apr 29, 2021