From: | "Jean-Christophe Arnu" <jcarnu(at)gmail(dot)com> |
---|---|
To: | "Koichi Suzuki" <koichi(dot)szk(at)gmail(dot)com> |
Cc: | pgsql-general(at)postgresql(dot)org |
Subject: | Re: lesslog "incorrect resource manager data checksum." |
Date: | 2008-11-20 15:48:35 |
Message-ID: | 5b3af220811200748h49456f78n2d69aa268e8c5b42@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-general |
Koichi-san,
full_page_writes are written to be on by default in the global user
configuration file (i.e. postgresql.conf). Query to the server showed that
it is :
show full_page_writes ;
full_page_writes
------------------
on
I use the RPM version provided on pgfoundry.
The main problem is located on pg_compresslog side (first) : it is unabe to
compress xlogs.
I send you (in private) a bunch of uncompressed logs that are known to
report pg_compresslog problem.
Once pg_compresslog will work, I plan to test "the other side" solution with
pg_decompresslog :)
Thanks for your reply!
2008/11/20 Koichi Suzuki <koichi(dot)szk(at)gmail(dot)com>
> Hi,
>
> Sorry I was out and late to find your report.
>
> Lesslog replaces full page write to corresponding incremental log to
> decrease the size and so you should turn full_page_writes "on" for
> lesslog to work.
>
> Did you specify "pg_decpmpresslog" as restore command? If so, I don't
> see anything missing so far. Pg_decompresslog must be used in place
> of "cat".
>
> If possible, could you try to share your archive log (a part may help)
> and let me know how you built lesslog? (Did you use rpm?) They will
> help. I've tested lesslog against pgbench and DBT-2 so I believe the
> coverage is reasonable.
>
> Sorry for the inconvenience and thanks a lot for your help.
>
> ---------
> Koichi Suzuki
>
> 2008/11/18 Jean-Christophe Arnu <jcarnu(at)gmail(dot)com>:
> > Hi,
> >
> > I'm trying to use lesslog 1.1 (out of NTT rpm available on pgfoundry) on
> my
> > pg_xlog files out of a 8.3.3 postgresql server.
> > I always get messages such as "incorrect resource manager data checksum."
> >
> > This message is issued (as I've read the sources) when there's CRC
> mismatch
> > in a record.
> >
> > Do you have any idea of what's going wrong (i.e. why my CRC are not
> > matching)? Does this deals with full_page_writes to be set to on (by
> default
> > in GUC) which is commented in my GUC (thus set to on) ? Is there
> something
> > missing?
> >
> > Note : I'm transferring my xlog to another node with rsync. The log I get
> (a
> > bunch of) for my test purpose are the ones on the remote node, they
> should
> > be exactly the same as the one on the server.
> >
> > Thanks by advance for any ideao!
> >
> > --
> > Jean-Christophe Arnu
> >
>
--
Jean-Christophe Arnu
From | Date | Subject | |
---|---|---|---|
Next Message | Richard Huxton | 2008-11-20 15:54:20 | Re: example of really weird caching (or whatever) problem |
Previous Message | Brandon Metcalf | 2008-11-20 15:34:25 | Re: example of really weird caching (or whatever) problem |