Hello Team, Is there a memory leak in compactor? I see compactor pod slowly accumulates memory and increase continuously, We have to restart the pods to get rid of this and this happens with only when pod when 4-5 replicas of compactor are running in parallel. We are running OSS helm chart version 0.92.0

Does all 4–5 replicas show this same climb, or just some of them ? Also, what are your

ZO_COMPACT_MAX_FILE_SIZE

and

ZO_MEMORY_CACHE_*

settings, if they're not the defaults .In the meantime, capping the memory cache on the compactor (

ZO_MEMORY_CACHE_MAX_SIZE

) or turning it off (

ZO_MEMORY_CACHE_ENABLED=false

) is a quick way to rule out the cache as the cause. cc: @Hengfei

we have got anyone report this before and ourself didn't find that. slowly increase, wasn't it OOM?

Does all 4–5 replicas show this same climb, or just some of them ?

just 1 pod the oldest of them

Also, what are your

ZO_COMPACT_MAX_FILE_SIZE

and

ZO_MEMORY_CACHE_*

settings, if they're not the defaults

They are the defaults Alert triggered when it crossed the requests and we recycled it before it hits the limits so no OOM.

which metrics are you looking for memory usage?

memory utilisation of compactor pod

what is the exactly metrics name