Just wondering if anyone else is seeing a slow increase in memory usage on 1.26.1 due to traceMemPool?
I have it enabled on my relay (8gb ram) which is my grafana host and I notice memory usage creep up from 58% as soon as the node’s restarted to 70% over a 24 hour period.
Not really an issue at the moment as my node is restarted every day for the topology update, but I thought this was meant to be resolved in 1.26.1?
yes, I can confirm
Just an update. When I ran a ledger-state query to check leaderlogs it crashed my node. My relay is an i5 with 8Gb Ram and 4gb Swap.
I’m increasing the swap for now until I can put another 8gb ram in there.
Are you planning to report this issue on cardano-node github site?
I can confirm as well. Since turning off traceMemPool, all of my nodes are much happier! No more dramatic memory usage increases or the fairly constant erratic dropping and reconnections of the nodes to the network. I had traceMemPool off on my relays for a couple of months, it was only this epoch that I turned it off on the BP’s because of posts I read from others that showed that my experience was not an isolated case.