Slow recovery tasks causing performance issues in ES cluster

Hi,

I noticed recently many recovery threads taking hours to run and the size of the data is quite small to cause this (MBs) what could be causing this ?

I also see huge skew gap of shard allocation across my cluster. It's currently in green status though.

GET _cat/recovery?v

Hi @viniciof
Which version are you using? Did you change any recovery settings?