We’ve considered that this might be related to a connection keepalive limit or something similar, as suggested. However, restarting the consumer doesn’t fix the issue. The only solution so far has been to restart my Ceramic node.
Could you tell more about the networking between the node and the client, please? I mean, there is a chance a proxy in front of a Ceramic node drops a connection.
Also, could you please elaborate on this statement: “And I’ve isolated the problem and tested the consumer by running it in the Ceramic container.” Does it mean you run the client and the node on the same container, and it fails the same?
Welcome @julissa and thank you very much for the super fast response!
@ukstv. Yep same container and no proxy between them.
It could actually be a file descriptor limit. I increased the limits and tried, but it didn’t fix the issue. Now I’m migrating to EKS. If it reproduces there, I’ll write back again.