One or more SHs in a SHC seem to go down frequently- Any suggestions?

[ksahu]

I have a SHC consisting of 4 SHs (Splunk on-prem on AWS). One or the other SHs seem to go into down state. The only info I can find is "Failed system status check" of the EC2 instance and splunk in stopped state. The splunkd.log amd health.log seems to be fine too.
Any suggestions which could solve the issue?

[isoutamo]

Hi

did I understood right, if I said, that you have 4 SH in SHC in your own AWS environment?

1st you should have odd number of SHs in SHC to get RAFT working correctly especially if you have those in several AZ/DC. To get majority after one (or more) node have gone down there must be more than 50% node left from original amount of nodes!

If you couldn't found anything from splunkd.log then you should also look messages and use dmesg -T command to check that e.g. omkiller haven't kill those on OoM (out of memory) situation? Also check what you have in your AWS CloudWatch etc.

r. Ismo