I've started monitoring certain things on our main database more proactively, and one thing I'm doing is tracking the _ActSummary table every 15 minutes or so. I noticed in promon that our buffers flushed stat was higher than I'd expected so wanted to track down when this is happening.
I've just looked at the report and at 19:00 2667 buffers are flushed. 19:00 also happens to be when the online backup started. Am I right in assuming that the backup is causing this, or do I need to look further? Is it a concern?
Progress 11.5.1 on Win 2012 R2.
We have 4 APWs. Yes 4. And at the moment we have 0 buffers flushed at any other time. I am trying to build a case for turning off a couple of these APWs and using their resources elsewhere!
A probkup online does flush modified buffers before it backs up the BI file. _ActSummary._Summary-Flushed shows buffers flushed *during a checkpoint*.
So... are the buffer flushes caused by probkup being counted in _Summary-Flushed, even if you're in the middle of a BI cluster? Don't know, I've never looked at that.
Four APWs sounds like a lot. Are you using -directio? What is your BI cluster size?
Is it possible the probkup online forces a checkpoint and hence the buffers are flushed at checkpoint?
Is BI Cluster Size the same as what some people call BI Block Size? If so we're at 8k and I want to move it to 16k. But I also want to move AI Block Size to 16k at the same time if possible which requires a re-source of replication.
Sorry not had enough coffee yet. BI Cluster Size is 32MB.
I would suggest applying the -Bp param with your online backup to prevent all buffers from being flushed out.
We have -Bp 64 set.
a checkpoint can begin only immediately after a bi cluster has become full and the previous checkpoint ended.
the backup program /does/ flush all modified buffer when it starts. thos get counted as buffers flushed because . . . flushes are flushes, regardless of the cause. a quiet point can also cause flushes. so does a database shutdown, but almost no one ever notices.
regards,
gus (gus@progress.com)
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." -- Brian Kernighan
> On Nov 26, 2015, at 5:54 PM, James Palmer wrote:
>
> Update from Progress Community [https://community.progress.com/]
>
> James Palmer [https://community.progress.com/members/jdpjamesp]
>
> I've started monitoring certain things on our main database more proactively, and one thing I'm doing is tracking the _ActSummary table every 15 minutes or so. I noticed in promon that our buffers flushed stat was higher than I'd expected so wanted to track down when this is happening.
>
> I've just looked at the report and at 19:00 2667 buffers are flushed. 19:00 also happens to be when the online backup started. Am I right in assuming that the backup is causing this, or do I need to look further? Is it a concern?
>
> Progress 11.5.1 on Win 2012 R2.
>
> We have 4 APWs. Yes 4. And at the moment we have 0 buffers flushed at any other time. I am trying to build a case for turning off a couple of these APWs and using their resources elsewhere!
>
> View online [https://community.progress.com/community_groups/openedge_rdbms/f/18/t/21635]
>
> You received this notification because you subscribed to the forum. To stop receiving updates from only this thread, go here [https://community.progress.com/community_groups/openedge_rdbms/f/18/t/21635/mute].
>
> Flag [https://community.progress.com/community_groups/openedge_rdbms/f/18/t/21635?AbuseContentId=b37abab5-2e8e-4657-ba9c-c6a42c29783b&AbuseContentTypeId=46448885-d0e6-4133-bbfb-f0cd7b0fd6f7&AbuseFlag=true] this post as spam/abuse.
Makes sense, Gus, but I thought that stat in _ActSummary was buffers flushed at checkpoint, rather than buffer flushed at any point. Or am I misunderstanding something (highly likely!)
there is no distinction and only one flushed counter.
regards,
gus (gus@progress.com)
"Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are,
by definition, not smart enough to debug it." -- Brian Kernighan
> On Nov 30, 2015, at 9:52 AM, James Palmer wrote:
>
> Update from Progress Community [https://community.progress.com/]
>
> James Palmer [https://community.progress.com/members/jdpjamesp]
>
> Makes sense, Gus, but I thought that stat in _ActSummary was buffers flushed at checkpoint, rather than buffer flushed at any point. Or am I misunderstanding something (highly likely!)
>
> View online [https://community.progress.com/community_groups/openedge_rdbms/f/18/p/21635/76041#76041]
>
> You received this notification because you subscribed to the forum. To unsubscribe from only this thread, go here [https://community.progress.com/community_groups/openedge_rdbms/f/18/t/21635/mute].
>
> Flag [https://community.progress.com/community_groups/openedge_rdbms/f/18/p/21635/76041?AbuseContentId=edf805a1-a097-48af-971f-a4613e497c29&AbuseContentTypeId=f586769b-0822-468a-b7f3-a94d480ed9b0&AbuseFlag=true] this post as spam/abuse.
Thanks Gus, in which case the schema definition of the field is a little misleading.
gyazo.com/9845a69d4c8922ad4b61025c2916e488
Not a major issue at all.
Anyway the upshot of this is that I will most likely be reducing my APW count at next restart!