Large garbage collection

[Barak]

Hi,

I am running constant indexing of ~500/s and ~3000/s reads on my cluster, and during that I see big drops in heap usage on the data nodes:

Cluster:
5 master nodes
7 data nodes
2 coordinator nodes

Is that an expected behavior or are there any changes that need to be done in regard of the garbage collection?

Thanks

[martinr_ubi (Martin R.)]

That's expected yes.

Please read those 4 documents:

Understanding the Memory Pressure Indicator

https://www.elastic.co/guide/en/cloud/current/ec-metrics-memory-pressure.html
https://www.elastic.co/blog/a-heap-of-trouble

Memory issues we'll remember

Tons of reading material but after the above I would guess you'd be at least in the 90th percentile in regards to "heap on ES". Complement as you go along or on an as needed basis.

[Barak]

Thanks @martinr_ubi !