filecache_max at 64%

[Lisa Dingman]

Our filecache_max is set to the default(16.3G or 32G), 50% but shows 96% usage. We have recently updated our licenses for our database product and made some changes to the udtconfig file for that. Not sure if this is the result of that or has been at 96% always. I will be able to reboot tonight to see if that makes a difference but could someone tell me what this is indicative of and what could be the possible problems with this being at 96% usage. Thank you in advance.

[Vivek Bhatia]

Hi Lisa,

Which version of HPUX?

How are you checking the settings of filecache_max? -Please paste the output in this forum

How are you checking the utilization of filcache_max ?

Regards
Vivek Bhatia

[James R. Ferguson · Accepted solution]

Hi Lisa:

I thing you mean that the filecache is at 96% of the 50% of physical memory that you apportioned (by default).

The 50% ceilig may or may not work well in your environment. As always, your mileage may vary. Do not equate the newer filecache parameters with the older buffer cache ones ('dbc_(min|max)_pct).

http://docs.hp.com/en/B2355-60130/filecache_max.5.html

Regards!

...JRF...

[Vivek Bhatia]

Hi,

Paste the output of kcusage and we might be able to understand the situation better.

Regards,
Vivek

[Steven E. Protter]

Shalom Lisa,

On any active system the filecache is going to get used, no matter the size.

The fact that its not 100% used, means its probably big enough.

If you reboot your system and check it after boot, you will find usage is lower than it is now.

This does not in itself indicate problems.

I am a member of the Bill Hassell school of administration. Problems get worked on when three is evidence, user complaints, slow response and such. I do many pro-active things like patching, but I don't get excited by a statistic like this.

SEP

[Lisa Dingman]

We have an integriy rx6600, hpux 11 iv3. Yes you are correct it shows 96% of the 50% of filecache_max. Just happened to see on graph in SMH.
Thank you for the advice about kcusage, am a programmer who is now is sysadmin so is new to me. This looks like it happened about the same time that the new udtconfig file for our database was updated, however they only suggested changing minimal amounts on fields, I have attached the changes requested.
# kcusage -d filecache_max
Tunable: filecache_max
Setting: 16322822144
Time Usage %
=============================================
Tue 06/23/09 11:00 CDT 4336123904 26.6
Tue 06/23/09 12:00 CDT 4386189312 26.9
Tue 06/23/09 13:00 CDT 4665380864 28.6
Tue 06/23/09 14:00 CDT 4702359552 28.8
Tue 06/23/09 15:00 CDT 4587450368 28.1
Tue 06/23/09 16:00 CDT 4661751808 28.6
Tue 06/23/09 17:00 CDT 4680110080 28.7
Tue 06/23/09 18:00 CDT 4686954496 28.7
Tue 06/23/09 19:00 CDT 4694245376 28.8
Tue 06/23/09 20:00 CDT 4700508160 28.8
Tue 06/23/09 21:00 CDT 4714946560 28.9
Tue 06/23/09 22:00 CDT 4811407360 29.5
Tue 06/23/09 23:00 CDT 4861390848 29.8
Wed 06/24/09 00:00 CDT 4864335872 29.8
Wed 06/24/09 01:00 CDT 4893179904 30.0
Wed 06/24/09 02:00 CDT 4827762688 29.6
Wed 06/24/09 03:00 CDT 16319885312 100.0
Wed 06/24/09 04:00 CDT 15383179264 94.2
Wed 06/24/09 05:00 CDT 15420329984 94.5
Wed 06/24/09 06:00 CDT 15306989568 93.8
Wed 06/24/09 07:00 CDT 15415889920 94.4
Wed 06/24/09 08:00 CDT 15671930880 96.0
Wed 06/24/09 09:00 CDT 15765381120 96.6
Wed 06/24/09 10:00 CDT 15817781248 96.9
#
I tried jus to see if anything changed to change back to the previous udtconfig file (for our datatel database) to see if that would make a difference because that is the only thing I can see that changed but so far it didn't:
kcusage -h filecache_max
Tunable: filecache_max
Setting: 16322822144
Time Usage %
=============================================
Wed 06/24/09 10:05 CDT 15784185856 96.7
Wed 06/24/09 10:10 CDT 15792558080 96.8
Wed 06/24/09 10:15 CDT 15793577984 96.8
Wed 06/24/09 10:20 CDT 15798099968 96.8
Wed 06/24/09 10:25 CDT 15808950272 96.9
Wed 06/24/09 10:30 CDT 15802822656 96.8
Wed 06/24/09 10:35 CDT 15811796992 96.9
Wed 06/24/09 10:40 CDT 15817834496 96.9
Wed 06/24/09 10:45 CDT 15831109632 97.0
Wed 06/24/09 10:50 CDT 15839285248 97.0
Wed 06/24/09 10:55 CDT 15839088640 97.0
Wed 06/24/09 11:00 CDT 15836065792 97.0

[Vivek Bhatia]

Hi Lisa,

How much is the physical Ram in your machine?

1. SHMMAX: 1073741824 1073741824 -This is currently 1GB and it should be equal to Physical RAM size.

Was there any changes on the server between 23/24?

Thanks
Vivek Bhatia

[Lisa Dingman]

We have 32G of RAM. We have web daemon processes that are allocated given amounts of memory so am not sure what upgrading this to 16G would do. The only change was that I saved the new udtconfig file, but that wouldn't take effect until the Database was restarted (4 AM this morning). I did load DataProtector on someone's pc yesterday at this time but I don't see how/if this would have effected this, maybe someone more knowlegeable than I see's a connection. That is the only thing that happened and it was exactly about 2 pm yesterday when the amount started climbing.

[Lisa Dingman]

Forgot to add, is there any problem with increasing shmmax? The man pages says to raise if it is below the max and user program are attempthing shmget segments ... and get EINVAL error message.(not happened) It shows 3% usage of the default of 1073741824, would that say the value is adequate since only at 3% usage?

[Vivek Bhatia]

Hi Lisa,

Well it is always recommended to have SHMMAX to be set equal to the Physical RAM.

you can run a kcusage to check how much this kernel parameter is in use?

Regards,
Vivek