added stopwaitsecs value of 300 seconds to supervisord programs

[JfeedZ]

In order to stop and resume pruned nodes with supervisord, geth needs more time to gracefully shut down. We can achieve this be setting stopwaitsecs in the supervisord.conf file.

If a value for stopwaitsecs isn't provided, the default value is 10 seconds. After this time has elapsed, supervisord will send a SIGKILL to any remaining child processes. For nodes that have synced a few million blocks, 10 seconds is not enough time for flushing to disk, and geth is stopped with SIGKILL. This results in incomplete/corrupt data, and the node will have to rewind and resync (often from genesis) when resumed. Increasing stopwaitsecs solves this issue by giving the appropriate amount of time for the blockchain to flush to disk before stopping. This will not increase the amount of time it takes to successfully stop a node, as supervisord doesn't wait the full duration once SIGCHLD is received.

[cb-heimdall]

Review Error for henridevieux @ 2024-02-27 18:13:03 UTC
User must have write permissions to review