Обсуждение: Postgres 9.0.4 + Hot Standby + FusionIO Drive + Performance => Query failed ERROR: catalog is missing 1 attribute(s) for relid 172226

Hi,

I tried with the PostgreSQL 9.0.4 + Hot Standby and running the database from Fusion IO Drive to understand the PG Performance.

While doing so I got the "Query failed ERROR: catalog is missing 1 attribute(s) for relid 172226". Any idea on this error? Is that combination PG + HotSB + Fusion IO Drive is not advisable?!

Regards,

Sethu Prasad. G.

> While doing so I got the "*Query failed ERROR: catalog is missing 1
> attribute(s) for relid 172226*". Any idea on this error? Is that combination
> PG + HotSB + Fusion IO Drive is not advisable?!

What were you doing when you got this error?

--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com

I did the hot standby configured earlier and at that time I started using(querying) the standby database.

May be something missed on the archive command.

On Tue, May 3, 2011 at 10:02 AM, Sethu Prasad <sethuprasad.in@gmail.com> wrote:

> I tried with the PostgreSQL 9.0.4 + Hot Standby and running the database
> from Fusion IO Drive to understand the PG Performance.
>
> While doing so I got the "Query failed ERROR: catalog is missing 1
> attribute(s) for relid 172226". Any idea on this error? Is that combination
> PG + HotSB + Fusion IO Drive is not advisable?!

Why I wonder do you think this might have anything to do with Hot
Standby and/or FusionIO drives?

This indicates either catalog or catalog index corruption of some kind.

Did you only get this error once?

--
 Simon Riggs                   http://www.2ndQuadrant.com/
 PostgreSQL Development, 24x7 Support, Training & Services

On 5/5/11 12:47 AM, Sethu Prasad wrote:
> I did the hot standby configured earlier and at that time I started
> using(querying) the standby database.
>
> May be something missed on the archive command.

Most likely, yes.  PostgreSQL shouldn't start up under such
circumstances, but apparently you fooled it.

--
Josh Berkus
PostgreSQL Experts Inc.
http://pgexperts.com

Yes it has something to do with Hot Standby, if you omit some parts on the archive then the standby instance will not have the necessary stuff and complain like this..

I kept the FusionIO drive in my checklist while attending to this issue, as we tried it looking for performance combined with read-only hot standby and in doubt I thought that the recovery is not successful on this drive safely. so I pointed that Fio Drive here.

Straight to say, I missed the pg_clog directory on archive.

seq_page_cost = 1.0

random_page_cost = 1.0
Is the above settings are fine when we deal with Fio and Performance, as I have the advice earlier stating that read and write are treated same with Fio drives.

Any suggestions on configuration changes to have read-only hot standby faster on READs.

- Sethu

On Tue, May 10, 2011 at 3:23 AM, Sethu Prasad <sethuprasad.in@gmail.com> wrote:
> Yes it has something to do with Hot Standby, if you omit some parts on the
> archive then the standby instance will not have the necessary stuff and
> complain like this..

If you omit some parts of the archive, it won't start at all.  To get
it to complain like this, you need something more than accidental
misconfiguration.

> I kept the FusionIO drive in my checklist while attending to this issue, as
> we tried it looking for performance combined with read-only hot standby and
> in doubt I thought that the recovery is not successful on this drive safely.
> so I pointed that Fio Drive here.
>
> Straight to say, I missed the pg_clog directory on archive.
>
> seq_page_cost = 1.0
>
> random_page_cost = 1.0
>
> Is the above settings are fine when we deal with Fio and Performance, as I
> have the advice earlier stating that read and write are treated same with
> Fio drives.

I would think more like 0.1 than 1.0.

> Any suggestions on configuration changes to have read-only hot standby
> faster on READs.

effective_io_concurrency?

Adjust OS readahead?

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company