Inquiry on hardcoded stream state flush timeout to 2 minutes

[souravagrawal]

Hello Team, I have been debugging an issue of stream state getting reset to 0 once NATS is abruptly killed (OOM). I am running single pod of NATS with a workqueue stream.
while debugging the issue I found if index.db is stale and blk file does not exist anymore it recover state from available blk files on reboot. There is a possibility of having no blk file as it might have failed to create one because it abruptly got killed.
Changing the stream state flush time from 2 minute to a lower could help (tested this), but its hardcoded.
I am trying to understand the cause of having it hardcoded and could it be made configurable.
I have also tried changing sync_interval to a lower value of 10sec that did not help too.
Here is a bug I filed with the steps to reproduce issue #6600

[derekcollison]

You can set the sync interval by adding the property under the jetstream configuration block for the server.

e.g.

jetstream {
  sync: 1s
}

[souravagrawal]

Hello @derekcollison I have gone through with syncblocks and tried changing it to lower value, it did not work. Changed it again with 1second but still did not fix the problem. I tried debugging the flow of syncblocks, It goes all the way to sync index.db fd but it has no impact.
Stream state still holds old data on abruptly terminating NATS.