Re: Help - corruption issue? - Mailing list pgsql-general

From Scott Marlowe
Subject Re: Help - corruption issue?
Date
Msg-id BANLkTimMDuaYR+hbWAuG-DGWFGj13F7HcA@mail.gmail.com
Whole thread Raw
In response to Re: Help - corruption issue?  (Phoenix Kiula <phoenix.kiula@gmail.com>)
Responses Re: Help - corruption issue?
List pgsql-general
On Mon, Apr 18, 2011 at 8:52 PM, Phoenix Kiula <phoenix.kiula@gmail.com> wrote:
> On Tue, Apr 19, 2011 at 8:35 AM, Scott Marlowe <scott.marlowe@gmail.com> wrote:
>> On Mon, Apr 18, 2011 at 5:44 PM, Tomas Vondra <tv@fuzzy.cz> wrote:
>>>
>>> Still, do the file backup as described in the previous posts. You could
>>> even do an online backup using pg_backup_start/pg_backup_stop etc.
>>
>> As soon as you have a working file system backup, get the tw_cli
>> utility for the 3ware cards downloaded and LOOK at what it has to say
>> about your RAID controller, drives, and array health.
>
>
>
> I am with SoftLayer. They're a very professional bunch. They even
> changed my BBU last night. The RAID card is working. The memory and
> the hardware are also tested.

So, RAID is good for sure?  As in someone logged into the machine, and
went to the tw_cli utility and asked it about the status of the
physical drives and virtual RAID array and the card said yes they're
good?  No bad sectors being remapped?  Hmmm.  One of my old tests when
things were acting up was to see if the server could compile the linux
kernel or pgsql back when it took 1.5 hours to do.  If you keep
getting sig 11s on production kernel compiles something's wrong with
the system, software or hardware.

> I have now upgraded to 8.2.19.
>
> Then I restarted the server, and dropped indexes. When I recreate the
> first index, the same thing happens:
>
> ------
> # CREATE INDEX idx_links_userid ON links (user_id);
> server closed the connection unexpectedly
>        This probably means the server terminated abnormally
>        before or while processing the request.
> The connection to the server was lost. Attempting reset: Failed.
> ------

What do the Postgresql logs say at this time?  oh wait...

> There is nothing going on in the server other than this command. All
> other users are blocked!
>
> Logging is enabled but does not have anything!

System logs maybe?  Something about a process getting killed?  Have
you tried turning up the verbosity of the pg logs?

> I am now worried. What is this problem?

We gotta check one thing at a time really.

If you copy the dir off to another machine and run pgsql 8.2.latest or
thereabouts, can you then create the index?

pgsql-general by date:

Previous
From: Steven Parkes
Date:
Subject: Re: replay doesn't catch up with receive on standby
Next
From: Phoenix Kiula
Date:
Subject: Re: Help - corruption issue?