Big slowdown on processing historical messages on reconnections

[matiwinnetou]

Hydra version: 0.9.0

We processed about 500 various messages, among transactions, a few SnapshotConfirms where UTxOs list was never longer than 1000 in size.

What we observe is that when websocket client disconnected and reconnected, the pattern is always the same. First there is nothing for 10, 20, 30 seconds (haven't exactly messured). CPU goes 100% for hydra node and then there is a BURST of all messages over websocket in one go.

This is now what we are expecting, when processing historical messages we expect messages to be gradually piped and without significant delay at the start.

[v0d1ch]

Hey @matiwinnetou 👋
Thanks for reporting this to us.

Do you have a way of reproducing this issue consistently and if so can you send us your local state and/or some instructions so we have easier time observing this issue on our end?

[matiwinnetou]

I think so, we can do some zoom session next week I think. I am using postman for validation so I also need to check with another client because could be postman issue as well.

[v0d1ch]

Hey @matiwinnetou any related news on this?

[v0d1ch]

@matiwinnetou can we close this issue or do you still think there is a problem?