pg.ddx.io  pgsql-admin@postgresql.org mailing list archive  
help / color / mirror / Atom feed
Block Read Error: Success?
2+ messages / 2 participants
[nested] [flat]

* Block Read Error: Success?
@ 2008-02-13 21:21  Thomas F. O'Connell <tf@o.ptimized.com>
  0 siblings, 1 reply; 2+ messages in thread

From: Thomas F. O'Connell @ 2008-02-13 21:21 UTC (permalink / raw)
  To: pgsql-admin

A few days ago, we began seeing this error in our logs:

ERROR:  could not read block [redacted] of relation [redacted]: Success

I've encountered "could not read block" errors in other databases in  
the past, but have never witnessed the "Success" tag at the end. What  
does that mean?

The error went unnoticed until, ultimately, significant parts of a  
database-driven website were not able to be read. Backups had been  
collected for days using pg_dump (which is configured to output SQL),  
and we were able to restore from a backup created earlier on the day  
that the content became unavailable. After restoration from backup,  
we haven't seen the errors again.

We're running 8.0.13 on FC3. Is there any reason not to suspect the  
hardware at this point?

--
Thomas F. O'Connell

optimizing modern web applications
: for search engines, for usability, and for performance :

http://o.ptimized.com/
615-260-0005

^ permalink  raw  reply  [nested|flat] 2+ messages in thread

* Re: Block Read Error: Success?
@ 2008-02-13 21:39  Tom Lane <tgl@sss.pgh.pa.us>
  parent: Thomas F. O'Connell <tf@o.ptimized.com>
  0 siblings, 0 replies; 2+ messages in thread

From: Tom Lane @ 2008-02-13 21:39 UTC (permalink / raw)
  To: Thomas F. O'Connell <tf@o.ptimized.com>; +Cc: pgsql-admin

"Thomas F. O'Connell" <tf@o.ptimized.com> writes:
> A few days ago, we began seeing this error in our logs:
> ERROR:  could not read block [redacted] of relation [redacted]: Success

> I've encountered "could not read block" errors in other databases in  
> the past, but have never witnessed the "Success" tag at the end. What  
> does that mean?

What this means is a read past EOF --- that is, read() returned fewer
bytes than requested (perhaps none at all) but did not signal an error.
A potential ultimate cause is corrupt indexes that contain pointers
leading past the end of the table.  Which sounds a whole lot like the
VACUUM bug fixed in 8.0.14 and other recent releases.

> We're running 8.0.13 on FC3. Is there any reason not to suspect the  
> hardware at this point?

I'd question your choice of platform --- FC3 went EOL quite some time
ago.

			regards, tom lane



^ permalink  raw  reply  [nested|flat] 2+ messages in thread


end of thread, other threads:[~2008-02-13 21:39 UTC | newest]

Thread overview: 2+ messages (download: mbox mbox.gz follow: Atom feed)
-- links below jump to the message on this page --
2008-02-13 21:21 Block Read Error: Success? Thomas F. O'Connell <tf@o.ptimized.com>
2008-02-13 21:39 ` Tom Lane <tgl@sss.pgh.pa.us>

This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox