Re: Kludge in pg_standby.c - Mailing list pgsql-hackers
| From | Bruce Momjian |
|---|---|
| Subject | Re: Kludge in pg_standby.c |
| Date | |
| Msg-id | 200804062354.m36NsbK21206@momjian.us Whole thread Raw |
| In response to | Kludge in pg_standby.c (Gregory Stark <stark@enterprisedb.com>) |
| Responses |
Re: Kludge in pg_standby.c
|
| List | 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@postgresql.org so that your
> message can get through to the mailing list cleanly
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://enterprisedb.com
+ If your life is a hard drive, Christ can be your backup. +
pgsql-hackers by date: