Files
neon/compute/etc/sql_exporter/pg_stats_userdb.sql
Alexey Kondratov 4b77807de9 fix(compute/sql_exporter): Ignore invalid DBs when collecting size (#11097)
## Problem

Original Slack discussion:
https://neondb.slack.com/archives/C04DGM6SMTM/p1739915430147169

TL;DR in Postgres, it's totally normal to have 'invalid' DBs (state
after the interrupted `DROP DATABASE`). Yet, some of our metrics
collected with `sql_exporter` try to get the size of such invalid DBs.

Typical log lines:
```
time=2025-03-05T16:30:32.368Z level=ERROR source=promhttp.go:52 msg="Error gathering metrics" error="[from Gatherer #1] [collector=neon_collector,query=pg_stats_userdb] pq: [NEON_SMGR] [reqid 0] could not read db size of db 173228 from page server at lsn 0/44A0E8C0"
time=2025-03-05T16:30:32.369Z level=ERROR source=promhttp.go:52 msg="Error gathering metrics" error="[from Gatherer #1] [collector=neon_collector,query=db_total_size] pq: [NEON_SMGR] [reqid 0] could not read db size of db 173228 from page server at lsn 0/44A0E8C0"
```

## Summary of changes

Ignore invalid DBs in these two metrics -- `pg_stats_userdb` and
`db_total_size`
2025-03-06 15:32:17 +00:00

21 lines
585 B
SQL

-- We export stats for 10 non-system databases. Without this limit it is too
-- easy to abuse the system by creating lots of databases.
SELECT pg_database_size(datname) AS db_size,
deadlocks,
tup_inserted AS inserted,
tup_updated AS updated,
tup_deleted AS deleted,
datname
FROM pg_stat_database
WHERE datname IN (
SELECT datname FROM pg_database
-- Ignore invalid databases, as we will likely have problems with
-- getting their size from the Pageserver.
WHERE datconnlimit != -2
AND datname <> 'postgres'
AND NOT datistemplate
ORDER BY oid
LIMIT 10
);