Golang hanging with Foundation 5.2.5

Development FoundationDB Layers
1

Hi,
I have found that the Golang driver can hang after some time when operating against FDB 5.2.5 but not in FDB 5.1.7.

Symptoms:

  • program executes correctly for a period of time then eventually becomes unresponsive (indicating some sort of memory leak).
  • happens when using watches against FDB but may not be related.
  • the FDB server is still fine as indicated by fdbcli status details and if the golang program is restarted everything behaves correctly.

I have been trying to write a minimal program to replicate the problem but unfortunately it takes time before the issue manifests so I haven’t been able to replicate yet.

I wanted to raise this issue here to see if anyone else is experiencing this issue.

As originally raised here

2

So after running some quite serious tests I have not been able to reproduce (perhaps I imagined it?)

  • 24 hours
  • 3 node FDB 5.2.5 cluster (configured double ssd-2) running on first gen intel NUC i3s.
  • 50 golang producers producing 20 versionstampedkey inserts/second of approx. 57 bytes (a timestamp) + updating a notify watch key
  • all 50 producers reading all messages based on the watch event with a 500ms rate limit on before each fetch
  • memory usage of around 30MB per producer stable
  • ability to do a hard reboot on any of the cluster nodes and everything just keeps running (amazing).
3

If you do manage to reproduce it and want to send along the client code that leads to the problem, I’d be happy to run it and see if I can figure out what’s going on.