Re: Kludge in pg_standby.c

From: Bruce Momjian <bruce(at)momjian(dot)us>
To: Gregory Stark <stark(at)enterprisedb(dot)com>, Magnus Hagander <magnus(at)hagander(dot)net>
Cc: pgsql-hackers list <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: Kludge in pg_standby.c
Date: 2008-04-06 23:54:37
Message-ID: 200804062354.m36NsbK21206@momjian.us
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers


Magnus, have you looked at this yet?

---------------------------------------------------------------------------

Gregory Stark wrote:
>
> There's a suspicious ifdef in pg_standby for WIN32 which smells like a kludge
> added to work around a Windows problem which makes it work but at great
> expense:
>
> #ifdef WIN32
> /*
> * Windows reports that the file has the right number of bytes
> * even though the file is still being copied and cannot be
> * opened by pg_standby yet. So we wait for sleeptime secs
> * before attempting to restore. If that is not enough, we
> * will rely on the retry/holdoff mechanism.
> */
> pg_usleep(sleeptime * 1000000L);
> #endif
>
> This happens before we return *any* WAL file to be processed. That means it
> slows down the processing of any file by 1s. On a server which has fallen
> behind this means it can't process files as quickly as it can copy them, it's
> limited to at most 1/s.
>
> I think it wouldn't be hard to do this properly. We can try to open the file,
> handle the expected Windows error by sleeping for 1s and repeating until we
> can successfully open it. Something like (untested):
>
> bool success = false;
> int fd, tries = 10;
> while (--tries) {
> fd = open(WALFilePath, O_RDONLY);
> if (fd >= 0) {
> close(fd);
> success = true;
> break;
> } else if (errno == EWINDOWSBLOWS) {
> usleep(1000000);
> } else {
> perror("pg_standby open:");
> exit(2);
> }
> }
> if (!success) {
> fprintf(stderr, "pg_standby: couldn't open file \"%s\" due to \"%s\",
> WALFilePath, strerror(EWINDOWSBLOWS));
> exit(2);
> }
>
>
> --
> Gregory Stark
> EnterpriseDB http://www.enterprisedb.com
> Ask me about EnterpriseDB's RemoteDBA services!
>
> ---------------------------(end of broadcast)---------------------------
> TIP 1: if posting/reading through Usenet, please send an appropriate
> subscribe-nomail command to majordomo(at)postgresql(dot)org so that your
> message can get through to the mailing list cleanly

--
Bruce Momjian <bruce(at)momjian(dot)us> http://momjian.us
EnterpriseDB http://enterprisedb.com

+ If your life is a hard drive, Christ can be your backup. +

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Tom Lane 2008-04-07 01:15:10 Re: Ordered Append Node
Previous Message Bruce Momjian 2008-04-06 23:53:51 Re: New style of hash join proposal