We recently setup netdata to monitor a self-hosted PostgreSQL server.
Is it a bad thing that txid_exhaustion_oldest_txid_num is constantly increasing?
What we did
PostgreSQL is running in v13
We've run a full-vacuum (
vacuumdb --full --all --jobs=1 --verbose --analyze --username=XXXX --password) on all logical databases and only got these errors:ERROR: relation "pg_temp_31.temp_lig_plo" ERROR: relation "pg_temp_31.lst_lig_plo"
We've dropped the schema "pg_temp_31" but still had a similar issue.
max_prepared_transactionsis 0SELECT * FROM pg_prepared_xactsis empty on all databasesWe've restarted the DB (so all idle transactions were disconnected) and still observe this increase

According to your comment, this graphic seems to be about transaction ID wraparound.
Check if any of your databases have a problem with that:
If that number is not much above 200 million for all databases, you don't need to worry. If the number is significantly higher, connect to that database and see which tables have a problem:
If there are any numbers far above 200 million (or whatever else
autovacuum_freeze_max_ageis set to), either autovacuum doesn't start on that table, or it is taking too long, or something is blocking its progress.Try processing those tables manually with
The output should tell you what is going on and, if possible, fix the problem.