We recently went to 7.6.5 and we started facing multiple performance issues...

- the cluster's shared disk went full... we had to manually stage the indexes to free up space.

- there would be multiple nsrindexd ADD processes and eventually backup would stall..

- we would need to restart networker in order to proceed with the backups..

- increased parallelism to 96 from 32.. have experienced a slight improvement...

Also observed that the Jobsdb purge was happening multiple times in a day......

Looking at the settings, the retention was at 3days/40MB (looks like that was the default)

Have changed the retention size x4 times as a trial run today to see if that helps... will keep u posted on the progress

However, would like to know if jobsdb purging frequently causes performance overheads that can stall backup completely !!! ???