|
Home > Archive > WebSphere Application Server > November 2007 > freepoolsize and poolsize pmi counters of jdbc connection pools
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
freepoolsize and poolsize pmi counters of jdbc connection pools
|
|
| johannes.kleinlercher@arz.co.at 2007-11-27, 7:37 am |
| Hi,<br />
<br />
we monitor the poolsize of our datasources with pmi.<br />
We retrieve a snapshot of the poolsize and freepoolsize counters every minute and I experienced very interesting data.<br />
Here is a part of the logfile we write:<br />
<br />
"Tue Nov 27 11:54:29 CET 2007" jdbc/ibd0g , PoolSize: 9 , FreePoolSize: 9 , "<br />
"Tue Nov 27 11:55:10 CET 2007" jdbc/ibd0g , PoolSize: 9 , FreePoolSize: 8 , "<br />
"Tue Nov 27 11:55:53 CET 2007" jdbc/ibd0g , PoolSize: 9 , FreePoolSize: 9 , "<br />
"Tue Nov 27 11:56:35 CET 2007" jdbc/ibd0g , PoolSize: 8 , FreePoolSize: 7 , "<br />
"Tue Nov 27 11:57:18 CET 2007" jdbc/ibd0g , PoolSize: 9 , FreePoolSize: 8 , "<br />
"Tue Nov 27 11:58:01 CET 2007" jdbc/ibd0g , PoolSize: 9 , FreePoolSize: 8 , "<br />
<br />
As you can see, the freepoolsize is always very high - actually it seems that most of the time the connections are in the freepool.<br />
Our settings of the datasource are as follows:<br />
<br />
minconnections=1<br />
maxconnections=50<br />
connectiontimeout=20 seconds<br />
unusedtimeout=120 seconds<br />
agedtimeout=0 seconds<br />
reaptime=60 seconds<br />
<br />
With these settings I assume that the freepoolsize should be very small because free connections should get destroyed after a maximum of 180 seconds (unusedtimeout + reaptime).<br />
<br />
Do you have any ideas why the freepoolsize is so high? Are there other parameters I missed?
| |
| Gopala Molakaluri 2007-11-27, 1:32 pm |
| If you are seeing this condition over a long time (during your Application activity window), it is likely that you are keeping atleast 9 of your connections active within 60-120secs. May be you have a transaction that is opening multiple connections in su
ccession and is releasing them at the end of the Tx. You probably are not noticing it in the log entries because your Tx is fast and is completing quickly. <br />
<br />
Once the Pool grows, I think Pool manager uses Round Robin alg, which may be another reason why you are growing up to 9 and staying there. <br />
<br />
-Dexthor.
| |
| johannes.kleinlercher@arz.co.at 2007-11-27, 1:32 pm |
| I understand what you mean.<br />
If one transaction needed 9 connections every minute and this transaction was very fast it would be likely that after the transaction the connections move in the freepool and never get destroyed because they get used within the usedtimeout again and again
. And it could be simply bad luck that I always take the snapshot when the connections are not in use ... <br />
<br />
Is this also your train of thoughts?<br />
<br />
Are there some better PMI statistics which tell me how utilized my connection pool is or should I just remain on the poolsize statistic and forget the freepoolsize?
|
|
|
|
|