IDS Forum

Re: RE: Long Checkpoints with HDR and 11.5.

Posted By: Art Kagel
Date: Monday, 15 March 2010, at 2:29 p.m.

In Response To: Re: RE: Long Checkpoints with HDR and 11.5. (JACQUES RENAUT)

No, Jacques, IDS secondaries do not share buffer pool updates with the
primary. Only the logical log buffers are transferred, so only the server's
transaction rate would affect the data transfer rate between the servers and
that would be about the same regardless of the LRU_MIN/MAX_DIRTY settings.
Now changing the size of the logical log buffer on the other hand, would
have some effect, but minor, and the discrepancy between the number of CPU
VPs on the two servers will have a much bigger effect.

Art

Art S. Kagel
Advanced DataTools (www.advancedatatools.com)
IIUG Board of Directors (art@iiug.org)

See you at the 2010 IIUG Informix Conference
April 25-28, 2010
Overland Park (Kansas City), KS
www.iiug.org/conf

Disclaimer: Please keep in mind that my own opinions are my own opinions and
do not reflect on my employer, Advanced DataTools, the IIUG, nor any other
organization with which I am associated either explicitly, implicitly, or by
inference. Neither do those opinions reflect those of other individuals
affiliated with any entity with which I am affiliated nor those of the
entities themselves.

On Mon, Mar 15, 2010 at 2:03 PM, JACQUES RENAUT <jrenaut@us.ibm.com> wrote:

> Ernie Knox wrote:
> We are receiving extreme slowness and long checkpoints on our 11.50.FC3
> and now 11.50.FC5 versions on AIX 6.1, while running HDR.
>
> Everything comes back to life when we cut HDR, but began again once we
> upgraded to FC5.
>
> Has anyone else had this issue and what was the resolution?
>
> IBM asked us to upgrade from FC3 to FC5, but that still doesn't seem to
> be the answer.
>
> Thanks,
> *******************************************************************
> Ernie Knox
> IT Database Administrator Specialist
> Sears Holdings
> 3333 Beverly Rd., B4-266A
> Hoffman Estates, IL. 60179
> Office: (847) 286-5735
> Email: Ernest.Knox@searshc.com
> Blackberry: 2244650553@messaging.sprintpcs.com
> Page via Skytel: 2244650553@sprint.skytel.com
> Informix or MySQL Primary: 9110210@skytel.com
> Informix or MySQL Secondary: 7276872@skytel.com
>
> " Yes we can make a Change! "
> " It's always a great day to watch Sports - GO LIONS, TIGERS, and BEARS!
> "
> " Lets not forget - GO Pistons and Red Wings! "
> GSU
> *******************************************************************
>
> Responce:
>
> Have you recently modified your LRU min/max parameters? It not, what do you
> have them set to on both the primary and secondary? If LRU max dirty is big
> (as you increased it because of the interval/non-blocking checkpoints),
> would
> you now be doing something on the primary that is actually causing your
> systems to reach that larger number of modified buffers prior to
> checkpointing? If so it could be the buffer pool flushing on the secondary
> as
> part of a checkpoint is done differently then the normal non-blocking
> buffer
> pool flushing on primaries. Because of this, if you have very active
> logical
> logging systems, you can back up your primary due to various mechanisms
> used
> on the seconadry. In FC5, if I remember correctly, we tried to help this
> out
> by detecting longer buffer pool flushes on the secondary and automatically
> reducing LRU min/max on the secondaries, but I don't recall how aggressive
> we
> were in reducing it, additionally, you'd have to hit at least 1 long
> checkpoint to get the LRU settings to take effect. In 11.50.xC6 we
> introduced
> logical log staging on secondaries during checkpoints (if enabled via the
> $ONCONFIG file) which I believe should completely negate the affect of the
> long checkpoint on the secondary from impacting the primary at all. Of
> course
> this is all based on the assumption that the long checkpoints are caused by
> a
> large amount of buffer flushing due to modifications to LRU min/max on the
> primary (which people started to do on 11.10 and 11.50 because of interval
> checkpoints).
>
> If/when you re-enable HDR I'd try to monitor the number of dirty buffers on
> your primary and secondary prior to checkpointing, especially if you have
> increased LRU min/max dirty (compared to what you would have been using
> back
> in pre 11.x days and blocking checkpoints).
>
> Jacques Renaut
> Informix Advanced Support
> APD Team
>
>
>
> *******************************************************************************
> Forum Note: Use "Reply" to post a response in the discussion forum.
>
>

--0015174c0f3ec920dc0481db0e51

Messages In This Thread

• RE: Long Checkpoints with HDR and 11.5.
  Knox, Ernest -- Monday, 15 March 2010, at 1:01 p.m.
  • RE: Long Checkpoints with HDR and 11.5.
    Madison Pruet -- Monday, 15 March 2010, at 1:07 p.m.
    • RE: Long Checkpoints with HDR and 11.5.
      Knox, Ernest -- Monday, 15 March 2010, at 1:41 p.m.
      • RE: Long Checkpoints with HDR and 11.5.
        Madison Pruet -- Monday, 15 March 2010, at 2:33 p.m.
  • Re: RE: Long Checkpoints with HDR and 11.5.
    JACQUES RENAUT -- Monday, 15 March 2010, at 2:03 p.m.
    • Re: RE: Long Checkpoints with HDR and 11.5.
      Art Kagel -- Monday, 15 March 2010, at 2:29 p.m.
      • Re: RE: Long Checkpoints with HDR and 11.5.
        JACQUES RENAUT -- Monday, 15 March 2010, at 3:07 p.m.
        • RE: RE: Long Checkpoints with HDR and 11.5.
          Knox, Ernest -- Tuesday, 16 March 2010, at 4:44 p.m.
          • Re: RE: RE: Long Checkpoints with HDR and 11.5.
            JACQUES RENAUT -- Tuesday, 16 March 2010, at 5:21 p.m.
            • Re: RE: RE: Long Checkpoints with HDR and 11.5.
              Madison Pruet -- Tuesday, 16 March 2010, at 7:23 p.m.
            • RE: RE: RE: Long Checkpoints with HDR and 11.5.
              Knox, Ernest -- Tuesday, 16 March 2010, at 11:44 p.m.
    • Re: RE: Long Checkpoints with HDR and 11.5.
      Madison Pruet -- Monday, 15 March 2010, at 2:34 p.m.

IDS Forum is maintained by Administrator with WebBBS 5.12.