threefoldtech/mycelium

Periodically reduce size of the seqno cache

LeeSmet opened this issue · 5 comments

While this might need some confirmation to originate from the seqno cache, v0.5.3 seems to consume a lot more memory on the public peers. Do some manual management there after removing items to reduce capacity if applicable.

v0.5.3 seems to consume a lot more memory on the public peers

How does the strategy to deploy new mycelium release? Maybe gradually each N hours or X days?

We generally deploy most of them at roughly the same time, but that shouldn't impact anything since we want stable memory consumption

After some more investigation it seems that it might not (entirely) be seqno cache related.

Adding some more debugging made it clear that this is, in fact, not caused by the seqno cache

Actual issue is probably related to quinn endpoints leaking. See #309