No space left on device
Hello! We are seeing this exception in our production environment: "could not resize shared memory segment "/PostgreSQL.154498794" to 4194304 bytes: No space left on device". We are confused that it refers to "device space" and are wondering in what kind of situation the Neon DB can run out of device space.
Is this "expected"? It started to happen suddenly without much change to our queries or data.
We have also scaled up the cluster from 4 CU to 7CU with no difference.
5 Replies
eastern-cyanOP•2y ago
Update: Scaling to 7CU did "fix" it, however it feels off that we have to use such a big cluster for the workload, especially since it worked before on much smaller instances.
subsequent-cyan•2y ago
@Tilon, we will have better metrics/observability in the future to help track these things down. I’m out at the moment so I can’t check dashboards. If this is affecting your production workload then I suggest opening a ticket using the support link in the console
eastern-cyanOP•2y ago
Thx, I opened a ticket 10 hours ago, however no response yet
still no response after 24h
like-gold•2y ago
Hey, did you get a response yet?
eastern-cyanOP•2y ago
yes, still a bit unclear what is causing it though. There seem to be some issues related to pgvector which could be a reason, but nothing conclusive so far.