[prev in list] [next in list] [prev in thread] [next in thread] 

List:       pgsql-bugs
Subject:    Re: [BUGS] BUG #7902: lazy cleanup of extraneous WAL files can cause out of disk issues
From:       Jeff Frost <jeff () pgexperts ! com>
Date:       2013-02-26 22:16:53
Message-ID: 512D3455.5040507 () pgexperts ! com
[Download RAW message or body]

On 02/26/13 09:08, Jeff Janes wrote:
>
>     They're all number of xlog based and the first one doesn't remove any,
>     that's not surprising,
>
>
> What was the state of the pg_xlog directory before you started the test?  If
> it were not full up to check_point segments * 2, (which is likely was not,
> since populating pgbench_accounts avoids WAL when not archiving), this I
> agree it is not surprising.

It always start at around 258 files.

The total WAL volume during the 10min pgbench run as evidenced by enabling an
archive command that just touches the filename is 1,972 WAL files archived, or
31G.  So, it is cleaning up or recycling about 9G, unfortunately, that's just
2G too few for a 20G filesystem.



-- 
Jeff Frost <jeff@pgexperts.com>
CTO, PostgreSQL Experts, Inc.
Phone: 1-888-PG-EXPRT x506
FAX: 415-762-5122
http://www.pgexperts.com/ 


[Attachment #3 (text/html)]

<html>
  <head>
    <meta content="text/html; charset=ISO-8859-1"
      http-equiv="Content-Type">
  </head>
  <body bgcolor="#FFFFFF" text="#000000">
    <div class="moz-cite-prefix">On 02/26/13 09:08, Jeff Janes wrote:<br>
    </div>
    <blockquote
cite="mid:CAMkU=1xnyJwrg4E7ia6JAnsA1QUkwSxL1upY_auoecM4b3Ubiw@mail.gmail.com"
      type="cite">
      <div><br>
      </div>
      <blockquote class="gmail_quote" style="margin:0 0 0
        .8ex;border-left:1px #ccc solid;padding-left:1ex">
        <div style="word-wrap:break-word">They're all number of xlog
          based and the first one doesn't remove any, that's not
          surprising,</div>
      </blockquote>
      <div><br>
      </div>
      <div>What was the state of the pg_xlog directory before you
        started the test? &nbsp;If it were not full up to check_point
        segments * 2, (which is likely was not, since populating
        pgbench_accounts avoids WAL when not archiving), this I agree it
        is not surprising.</div>
    </blockquote>
    <br>
    It always start at around 258 files.<br>
    <br>
    The total WAL volume during the 10min pgbench run as evidenced by
    enabling an archive command that just touches the filename is 1,972
    WAL files archived, or 31G.&nbsp; So, it is cleaning up or recycling
    about 9G, unfortunately, that's just 2G too few for a 20G
    filesystem.<br>
    <br>
    <br>
    <br>
    <pre class="moz-signature" cols="78">-- 
Jeff Frost <a class="moz-txt-link-rfc2396E" \
href="mailto:jeff@pgexperts.com">&lt;jeff@pgexperts.com&gt;</a> CTO, PostgreSQL \
                Experts, Inc.
Phone: 1-888-PG-EXPRT x506
FAX: 415-762-5122
<a class="moz-txt-link-freetext" \
href="http://www.pgexperts.com/">http://www.pgexperts.com/</a>  </pre>
  </body>
</html>



[prev in list] [next in list] [prev in thread] [next in thread] 

Configure | About | News | Add a list | Sponsored by KoreLogic