Open etherscoredao opened 2 weeks ago
I guess your cluster / stack / compose / start command is interesting here as you can limit resources there.
For myself I limited the containers in my swarm because I had no luck doing it with the node's config. I also switched to ipns -> dhtclient to reduce traffic on some nodes. Maybe this also reduces the memory footprint (wild guess).
It seems that, the problem is introduced since 0.28.0, and has been fix in the master branch. So I would revert to 0.27.0 .
Checklist
Installation method
ipfs-desktop
Version
Config
Description
Hi, memory usage increases endlessly. The ipfs kubo node is installed on a server with:
I use docker stats to monitor the memory usage, here are the current log:
Debugging infos: memory stack: ipfs.stack.tar.gz memory heap: ipfs.heap.tar.gz cpu profile: ipfs.cpuprof.tar.gz
Looks like some goroutines are hanging like:
more infos in the attached memory stack file