We are exploring CockroachDB in a 3 node Kubernetes cluster. Each has 2 CPU, 3GB memory.
I’m very confused with DELETE statement performance and probably doing something wrong.
On a table of about 7 millions record it can take an hour to delete 25K rows (attempt on the screenshot below). Biggest try was 5 hours running statement after which I just restarted the cluster.
What seems off to me is why one statement on idle cluster can’t manage its execution without locks and retries. And why there’s no limit on number of automatic retries.
I have a trace to share https://gist.github.com/acnot/d4f4a039fc066deb125c46b75afc02b7
This time it took 236840.902ms to delete 10K rows. In the trace you can see a lot of locks and retries.
Is this normal behavior for CockroachDB? Anything I’m doing wrong?