Memory keeps increasing

[RicoToothless]

We have 40 Kusama validators and host on EKS cluster.

The ValidatorOfflineSessionLong metric start inaccurate and trigger the alert. It was recovered by itself in a short time until it was totally unrecoverable.

But our Kusama validators and we run Polkadot watcher locally to test the metrics were fine.

We found out the Polkadot watcher memory usage keeps increasing.

After recreating the pod, Polkadot watcher was fine for a while.

[ironoa]

Thanks for noticing that, I'll have a look at this