On Wed, Jan 24, 2024 at 1:05 PM Nathan Bossart <nathandbossart@gmail.com> wrote:
> There might be an overflow risk in the cutoff time calculation, but I doubt
> that's the root cause of these failures:
>
> /*
> * Files should only be removed if the last modification time precedes the
> * cutoff time we compute here.
> */
> cutoff_time = time(NULL) - 60 * wal_summary_keep_time;
>
> Otherwise, I think we'll probably need to add some additional logging to
> figure out what is happening...
Where, though? I suppose we could:
1. Change the server code so that it logs each WAL summary file
removed at a log level high enough to show up in the test logs.
2. Change the TAP test so that it prints out readdir(WAL summary
directory) at various points in the test.
--
Robert Haas
EDB: http://www.enterprisedb.com