Обсуждение: wal archiving question
Hi all -
We have postgres 9.0.4 since an year. Replication also set up and works fine. On the master. The wal files are growing and it's almost gone to 90% of the disk. How can I handle the situation. Can I delete the old files. Please advice
Regards
Thanks for the help
We have postgres 9.0.4 since an year. Replication also set up and works fine. On the master. The wal files are growing and it's almost gone to 90% of the disk. How can I handle the situation. Can I delete the old files. Please advice
Regards
Thanks for the help
akp geek wrote: > We have postgres 9.0.4 since an year. Replication also set up and > works fine. On the master. The wal files are growing and it's > almost gone to 90% of the disk. How can I handle the situation. > Can I delete the old files. Please advice Replicating how? Are these WAL files in the pg_xlog directory, the location your archive script copies them to, or somewhere else? What is your overall backup strategy? How many base backups do you keep, how far back? What sort of recovery options do you need (e.g., point in time recovery to any point in the last year, or something less extreme?) -Kevin
akp geek wrote: > We have postgres 9.0.4 since an year. Replication also set up and works fine. > On the master. The wal files are growing and it's almost gone to 90% of the disk. > How can I handle the situation. Can I delete the old files. Please advice Are the WAL files (the ones in pg_xlog) growing or the WAL archives? If the former, then you maybe got an archiving problem (consult the server log). If the latter, you'd have to manually delete some, but take care that you don't delete any that you still need either for restoring a backup or for the standby to catch up. Yours, Laurenz Albe
WAL archives is growing, not the files in pg_xlog.
Regards
Regards
On Wed, Dec 5, 2012 at 3:38 PM, Albe Laurenz <laurenz.albe@wien.gv.at> wrote:
akp geek wrote:Are the WAL files (the ones in pg_xlog) growing or the WAL archives?
> We have postgres 9.0.4 since an year. Replication also set up and works fine.
> On the master. The wal files are growing and it's almost gone to 90% of the disk.
> How can I handle the situation. Can I delete the old files. Please advice
If the former, then you maybe got an archiving problem
(consult the server log).
If the latter, you'd have to manually delete some,
but take care that you don't delete any that you still
need either for restoring a backup or for the standby to
catch up.
Yours,
Laurenz Albe
akp geek <akpgeek@gmail.com> writes: > WAL archives is growing, not the files in pg_xlog. You need to prune the archives of files that are too old to be interesting. This doesn't happen automatically since PG has no idea what your archiving requirements are. Be sure to keep WAL files at least back to the start of your last base backup. Some people like to keep them back to the next-to-last base backup. regards, tom lane
thank you all. There is a mistake I did then. I created a base backup when I started the DB , that was exactly an year ago. I need to do that again and then delete the old files.
Regards
Regards
On Wed, Dec 5, 2012 at 4:25 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
You need to prune the archives of files that are too old to be
interesting. This doesn't happen automatically since PG has no
idea what your archiving requirements are.
Be sure to keep WAL files at least back to the start of your
last base backup. Some people like to keep them back to the
next-to-last base backup.
regards, tom lane
akp geek <akpgeek@gmail.com> writes: > thank you all. There is a mistake I did then. I created a base backup when > I started the DB , that was exactly an year ago. I need to do that again > and then delete the old files. Yeah, you should be making new base backups a lot more frequently than that, probably. Consider: if your DB died today and you had to restore from the base backup, you'd have to replay a year's worth of WAL to get back where you were. That'd take ages. So the bloat in the archive directory is just one reason not to do it like that. regards, tom lane
thank you so much for all the inputs.. One final question is , Do we have to stop the database when we create the base backup?
Regards
Regards
On Wed, Dec 5, 2012 at 4:45 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
akp geek <akpgeek@gmail.com> writes:Yeah, you should be making new base backups a lot more frequently than
> thank you all. There is a mistake I did then. I created a base backup when
> I started the DB , that was exactly an year ago. I need to do that again
> and then delete the old files.
that, probably. Consider: if your DB died today and you had to restore
from the base backup, you'd have to replay a year's worth of WAL to get
back where you were. That'd take ages. So the bloat in the archive
directory is just one reason not to do it like that.
regards, tom lane
akp geek wrote: > thank you so much for all the inputs.. One final question is , Do we have to stop the database when we > create the base backup? No, see http://www.postgresql.org/docs/9.0/static/continuous-archiving.html#BACKUP-BASE-BACKUP Yours, Laurenz Albe