sql,obs: metrics / logs that break down SQL errors by group or type about cockroach HOT 6 OPEN

We will add metrics that count errors based on the "class", which is the first 2 letters of the error code: https://www.postgresql.org/docs/current/errcodes-appendix.html

from cockroach.

Did we look at the failed execution insights? What's missing here?

https://www.cockroachlabs.com/docs/stable/ui-insights-page#failed-execution

This table's contents are also conveniently dumped into debug zip.

from cockroach.

Looking at the internal thread a bit more closely, seems like we'd prefer an easier process than using debug zip and consume this information via metrics?

from cockroach.

It would also be nice to have a log like cockroach-sql-slow that records all failed statement executions (instead of slow executions) and is opt-out rather than opt-in.

(The reason I think we also need a log is that it's going to be difficult to dig into XXUUU errors without something more specific than a single count for this class.)

EDIT: I guess this shows up in cockroach-sql-slow anyway but it would be nice if logging of failed executions was opt-out rather than opt-in.

from cockroach.

Hi, I'm interested in taking on this issue. Could you give me more details?

from cockroach.

EDIT: I guess this shows up in cockroach-sql-slow anyway but it would be nice if logging of failed executions was opt-out rather than opt-in.

I think your request still stands -- cockroach-sql-slow will only have queries that take longer than the slow query threshold, but won't have queries that failed quickly.

from cockroach.