[chart/redis-ha>][BUG]

[athill]

Describe the bug
Updating to chart > 3.35.3 causes a spike in CPU usage

To Reproduce
Steps to reproduce the behavior:

1. Update the chart to 3.35.4 or above, using any version of Redis
2. Deploy changes
3. CPU spikes

For example in a 3 node cluster, going from 3.35.3 to 3.25.5 CPU usage:
Node 1: 156 -> 228
Node 2: 189 -> 245
Node 3: 161 -> 180

Expected behavior
The CPU usage should remain stable

Additional context
These numbers do not diminish after the the upgrade is complete and the cluster stabilizes. This is also a cluster that receives virtually no traffic.

[DandyDeveloper]

@athill Which container is taking this spike in CPU? If its Redis, it won't be the chart and likely might be something in the versioning?

If its the sidecars, can you get some monitoring attached and see if there's a specific container in the stack that has triggered this?

[athill]

@DandyDeveloper

First. let me thank you for your work on maintaining this chart. I forgot to mention that previously.

I also forgot to mention that the issue appears to be in the split-brain-fix container. Before upgrading to >4.35.3 in a cluster with limited usage, its CPU usage was 0; after upgrading, its cpu is averaging around 70.

I hope this helps.

Thank you.

[DandyDeveloper]

@athill This'll take a bit of time to check, but I'll use the latest chart as a baseline and see why / where the split brain might be excessively called.

There has been a number of updates to that script lately to solidify the logic. It's very possible these have caused it to run more regularly, and as a result, consume more CPU.

[athill]

@DandyDeveloper Thank you! No hurry on our end. 3.35.3 is working fine for now.