Re: O(n^2) system calls in RemoveOldXlogFiles()

Hi,

On 2021-01-11 16:35:56 +1300, Thomas Munro wrote:
> I noticed that RemoveXlogFile() has this code:
>
> /*
> * Before deleting the file, see if it can be recycled as a future log
> * segment. Only recycle normal files, pg_standby for example can create
> * symbolic links pointing to a separate archive directory.
> */
> if (wal_recycle &&
> endlogSegNo <= recycleSegNo &&
> lstat(path, &statbuf) == 0 && S_ISREG(statbuf.st_mode) &&
> InstallXLogFileSegment(&endlogSegNo, path,
> true,
> recycleSegNo, true))
> {
> ereport(DEBUG2,
> (errmsg("recycled write-ahead log file \"%s\"",
> segname)));
> CheckpointStats.ckpt_segs_recycled++;
> /* Needn't recheck that slot on future iterations */
> endlogSegNo++;
> }
>
> I didn't check the migration history of this code but it seems that
> endlogSegNo doesn't currently have the right scoping to achieve the
> goal of that last comment, so checkpoints finish up repeatedly search
> for the next free slot, starting at the low end each time, like so:
>
> stat("pg_wal/00000001000000000000004F", {st_mode=S_IFREG|0600,
> st_size=16777216, ...}) = 0
> ...
> stat("pg_wal/000000010000000000000073", 0x7fff98b9e060) = -1 ENOENT
> (No such file or directory)
>
> stat("pg_wal/00000001000000000000004F", {st_mode=S_IFREG|0600,
> st_size=16777216, ...}) = 0
> ...
> stat("pg_wal/000000010000000000000074", 0x7fff98b9e060) = -1 ENOENT
> (No such file or directory)
>
> ... and so on until we've recycled all our recyclable segments. Ouch.

I found this before as well: https://postgr.es/m/CAB7nPqTB3VcKSSrW2Qj59tYYR2H4+n=5pZbdWou+X9iqVNMCag@mail.gmail.com

I did put a hastily rebased version of that commit in my aio branch
during development: https://github.com/anarazel/postgres/commit/b3cc8adacf7860add8cc62ec373ac955d9d12992

Greetings,

Andres Freund