Every week we have to restart jenkins due to OOM. Continuously Metaspace increases and the number of classes loaded are also increasing.

Earlier we had this problem with both master and windows slaves. But we figured out that disabling the JNLP3 protocol on windows slave and not running groovy script on slaves solved the problem. But. it is still a problem on master and restarting a master is not anyone is expecting.

We recently upgraded our jenkins environment to 2.7.2 and took all the latest plugins.
We have moved all our flow jobs to pipeline except one.