Open SgtPooki opened 1 year ago
Note that I related #3320 because this type of behavior could easily kill routers... but it did not kill mine.
How many other calls are we sending without exponential backoff and circuit breaking on failures?
Do we not have exponential backoff and/or circuit breaking? Why is nat/nat.go:179 called SO many times per minute? It looks like we have some throttling, https://github.com/ipfs/kubo/blob/88d431c8129aeef2ea5076336412cb54161342ce/core/node/libp2p/nat.go#L20, but that appears to be on an individual basis and that there are multiple consumers/deps trying to use AutoNatService.
Checklist
Installation method
built from source
Version
Config
Description
Ran my kubo daemon with
GOLOG_LOG_LEVEL="warn,webtransport=debug,reuseport-transport=warn,net/identify=error" cmd/ipfs/ipfs daemon 2>&1 | tee kubo-log.txt
because I was debugging other stuff. I left it running since friday, and my laptop slept over the weekend. When opening it back up, I have a spamming of these errors:full log at https://bafybeiaxuasctnq3sgwwgtgdk6q2rlvuwe4s37bfzin7z7cnt6byesdpvq.ipfs.w3s.link/bafybeig6sodjrftg5eetjzlcvm7nx4qcwgwbe72n4ps6nuebdemj5aa62u
Repro?
Not yet.. i'll add to this if it occurs after a long sleep again
Killing daemon
CTRL+C
and restarting it does not show me these errors.Other Details
My public IP (dynamic IP from ISP) is different than what it was on Friday
Clear bugs (to me)
Related issues
Couldn't find any in this repo, but I found the following in lotus, which I don't think actually addresses the root issue:
Potentially related issues