Custom query with pg_database_size causes 500 after upgrading from 0.10.1 to 0.11.1

[dastrobu]

What did you do?

I upgraded from 0.10.1 to 0.11.1.

What did you expect to see?

the exporter works unchnged

What did you see instead? Under which circumstances?

the metrics endpoint fails with 500

Environment

• System information:

  Linux 3.10.0-1160.59.1.el7.x86_64 x86_64

• postgres_exporter version:

postgres_exporter, version 0.11.1 (branch: HEAD, revision: b5fd2465646a09bb1c8dbd5c9c00b852da93927a)
  build user:       root@a2c981f603ff
  build date:       20220818-20:58:17
  go version:       go1.18.5
  platform:         linux/amd64

• postgres_exporter flags:

--log.format=json --log.level=debug

• PostgreSQL version:

  12.7.0

• Logs:

  | {"address":":9187","caller":"main.go:135","level":"info","msg":"Listening on address","ts":"2022-09-21T14:54:28.486Z"}
-- | --
  | {"caller":"tls_config.go:195","http2":false,"level":"info","msg":"TLS is disabled.","ts":"2022-09-21T14:54:28.492Z"}
  | {"caller":"server.go:74","fingerprint":"postgresql:5432","level":"info","msg":"Established new database connection","ts":"2022-09-21T14:55:00.262Z"}
  | {"caller":"postgres_exporter.go:617","level":"debug","msg":"Querying PostgreSQL version","server":"postgresql:5432","ts":"2022-09-21T14:55:00.388Z"}
  | {"caller":"postgres_exporter.go:645","from":"0.0.0","level":"info","msg":"Semantic version changed","server":"postgresql:5432","to":"12.7.0","ts":"2022-09-21T14:55:00.399Z"}
  | {"caller":"collector.go:200","duration_seconds":0.417803329,"level":"debug","msg":"collector succeeded","name":"database","ts":"2022-09-21T14:55:00.682Z"}
  | {"caller":"collector.go:200","duration_seconds":0.320640151,"level":"debug","msg":"collector succeeded","name":"bgwriter","ts":"2022-09-21T14:55:00.678Z"}
  | {"cache_seconds":30,"caller":"queries.go:213","level":"debug","metric":"pg_database","msg":"New user metric namespace from YAML metric","ts":"2022-09-21T14:55:00.883Z"}
  | {"cache_seconds":30,"caller":"queries.go:213","level":"debug","metric":"pg_stat_user_tables","msg":"New user metric namespace from YAML metric","ts":"2022-09-21T14:55:00.883Z"}
  | {"caller":"queries.go:267","level":"debug","metric":"pg_database","msg":"Adding new metric from user YAML file","ts":"2022-09-21T14:55:00.883Z"}
  | {"caller":"queries.go:267","level":"debug","metric":"pg_stat_user_tables","msg":"Adding new metric from user YAML file","ts":"2022-09-21T14:55:00.883Z"}
  | {"caller":"queries.go:278","level":"debug","msg":"Adding new query override from user YAML file","query_override":"pg_database","ts":"2022-09-21T14:55:00.883Z"}
  | {"caller":"queries.go:278","level":"debug","msg":"Adding new query override from user YAML file","query_override":"pg_stat_user_tables","ts":"2022-09-21T14:55:00.883Z"}
  | {"caller":"pg_setting.go:35","level":"debug","msg":"Querying pg_setting view","server":"postgresql:5432","ts":"2022-09-21T14:55:00.887Z"}
  | {"caller":"namespace.go:193","level":"debug","msg":"Querying namespace","namespace":"pg_database","ts":"2022-09-21T14:55:01.083Z"}
  | {"caller":"namespace.go:193","level":"debug","msg":"Querying namespace","namespace":"pg_stat_user_tables","ts":"2022-09-21T14:55:01.087Z"}

The query, which seems to cause the issue is as follows:

    queries:
      pg_database:
        query: "SELECT pg_database.datname, pg_database_size(pg_database.datname) as size_bytes FROM pg_database;"
        master: true
        cache_seconds: 30
        metrics:
          - datname:
              usage: "LABEL"
              description: "Name of the database"
          - size_bytes:
              usage: "GAUGE"
              description: "Disk space used by the database"

without the pg_database_size(pg_database.datname) it seems to work fine.

When downgrading again to version 0.10.1, everything works as expected.

[sysadmind]

#613 Added the pg_database metrics into the exporter itself. You don't need it in the queries file with 0.11.0+