PostgreSQL was configured to wrote WAL archive into NFS server. After some time, suddenly, WAL writing stopped. In the log:
2026-01-23 18:46:49.970 MSK,,,12085,,697397e9.2f35,1,,2026-01-23 18:46:49 MSK,,0,WARNING,22023,"invalid value for parameter ""basic_archive.archive_directory"": ""/mnt/ocean/postgres/stars/WAL""","Specified archive directory does not exist.",,,,,,,,"","archiver",,0
2026-01-23 18:57:23.589 MSK,,,12085,,697397e9.2f35,2,,2026-01-23 18:46:49 MSK,,0,WARNING,01000,"""archive_mode"" enabled, yet archiving is not configured","basic_archive.archive_directory is not set.",,,,,,,,"","archiver",,0
And the last row repeated several days, pg_wal raised up to 4 TiB. I checked, the basic_archive.archive_directory was configured:
postgres=# show basic_archive.archive_directory;
basic_archive.archive_directory
-----------------------------------------
/mnt/ocean/postgres/stars/WAL
(1 row)
And directory existed and it has old WAL files (before error raised).
The basic_archive is unstable, but I don't know how to repeat this. Workaround: use the oldschool archive_command .