Re: some error messages in postgeslog

surabhi.ahuja wrote:
> I use PostgreSQL 8.0.0
>
> it seems that the disk was close to full,
>
> i executed a program (in C++) which opens a connection to Postgres using PQConnectdb.
>
> and then it reads from a particluar table, and simply displays the values on the console.
>
> after doing that it will close the connection using PQfinish
>
> and thats when these error messages come....

> <2006-04-05 17:10:47 CDT%>DETAIL: could not write to file "/export/home1/sdc_image_pool/dbx/pg_clog/0001" at offset 40960:
> No space left on device
> <2006-04-05 17:10:48 CDT%>LOG: could not close temporary statistics file "/export/home1/sdc_image_pool/dbx/global/pgstat.tm
> p.24783": No space left on device
> <2006-04-05 17:10:49 CDT%>ERROR: could not access status of transaction 0
> <2006-04-05 17:10:49 CDT%>DETAIL: could not write to file "/export/home1/sdc_image_pool/dbx/pg_clog/0001" at offset 40960:
> No space left on device
> <2006-04-05 17:10:50 CDT%>ERROR: could not access status of transaction 0
> <2006-04-05 17:10:50 CDT%>DETAIL: could not write to file "/export/home1/sdc_image_pool/dbx/pg_clog/0001" at offset 40960:
> No space left on device

> is it true that for starting a connection to postgres read a table and then closing connection disk space will be needed.

It looks like it's the statistics monitoring that initially causes a
problem here, so yes. You would in any case need some disk-space I'm
sure, not necessarily a lot though.

> Is it because of this reason these error messages are coming?

Yes, the "No space left on device" error does imply that you have no
space left on that disk device. If this is so, you will need to add more
storage or delete something else from it.

> i have another question i have also noticed this error message in the postgres log, what are its implications and what happened that this error message came
> :
> 2006-04-05 09:22:08 CDT%>LOG: received fast shutdown request
> <2006-04-05 09:22:08 CDT%>LOG: aborting any active transactions
> <2006-04-05 09:22:08 CDT%>LOG: checkpoints are occurring too frequently (16 seconds apart)
> <2006-04-05 09:22:08 CDT%>HINT: Consider increasing the configuration parameter "checkpoint_segments".

I don't see the word "error" anywhere in these log lines. Not everything
logged is an error.

Checkpoints are to do with the Write-Ahead-Log (WAL) which logs
transaction details before the changes are written to disk. They occur
after a pre-defined time or when you reach the limit of
"checkpoint_segments" wal-files written. You can find out more in the
manuals.
http://www.postgresql.org/docs/8.1/static/wal.html
http://www.postgresql.org/docs/8.1/static/runtime-config-wal.html

HTH
--
Richard Huxton
Archonet Ltd