radicle-org-node gets stuck, ignoring traffic, spinning 2 cpu cores

[kermit4]

netstat shows unread packets accumulating
top shows 200% CPU usage

the last line in logs was:

Aug 19 19:18:34.619 INFO ureq::unit: sending request POST https://api.thegraph.com/subgraphs/name/radicle-dev/radicle-orgs

this is at commit 0a4a395
release build

Linux 4.18.0-305.3.1.el8.x86_64
CentOS Linux release 8.4.2105

The CPU usage may be unrelated as it does that at times it is responding to network traffic.

@kim thinks it may be due to radicle-dev/radicle-link#734

[kim]

Oh so this is a radicle org node. Would capturing logs not suggest that #5, radicle-dev/radicle-link#734 and this one are related?

[cloudhead]

Yeah, they are all the same issue AFAICT.

[cloudhead]

I'd like to upgrade the seed node to the latest radicle-link, as well as the org node, and see if this is still an issue 🤔

[kermit4]

I think it stopped leaking memory around the same time it began using a lot of CPU so I think #5 is fixed.

There's not much network traffic so I don't think it relates to radicle-dev/radicle-link#734 either.

[kermit4]

For the hangs (probably unrelated to CPU usage) In at least 3 of 3 cases, the last line in logs is:

INFO ureq::unit: sending request POST https://api.thegraph.com/subgraphs/name/radicle-dev/radicle-orgs

Maybe that request isn't spawning its own thread?

[kermit4]

Dropping traffic to api.thegraph.com didn't cause hangs so maybe not. Maybe that's just the part that wasn't hung so is why it kept showing up in logs when nothing else was.

[kermit4]

For 30 minutes while the network receive queue was full, until I restarted it, "INFO ureq::unit: sending request POST https://api.thegraph.com/subgraphs/name/radicle-dev/radicle-orgs" was the only line in the logs.

[cloudhead]

So what you're seeing now is:

• Normal memory consumption
• High CPU usage
• Normal network usage

Is that right?

[sebastinez]

Closing this issue since we deprecated the org-node some time ago.