IDS Forum

Re: PC_POOLSIZE?

Posted By: JACQUES RENAUT
Date: Wednesday, 21 August 2019, at 9:15 p.m.

In Response To: PC_POOLSIZE? (DAVID GROVE)

Original post:

IDS 12.10.FC12
Solaris 10 1/13

How does PC_POOLSIZE get changed?

In our ONCONFIG file, the PC_POOLSIZE parameter is entered as:

PC_POOLSIZE 127

The first few lines of 'onstat -g prc' are as follows:

informix@ifmx-prod-anc>onstat -g prc|more
IBM Informix Dynamic Server Version 12.10.FC12 -- On-Line -- Up 22 days 08:15:40 -- 19693568 Kbytes
UDR Cache:

Number of lists : 31

PC_POOLSIZE : 566

How did PC_POOLSIZE get to be 566?

I do have AUTO_TUNE set to 1, but I don't seem to see anything in the documentation for "AUTO_TUNE" that would lead me to believe that PC_POOLSIZE is one of the parameters that AUTO_TUNE adjusts dynmaically. Maybe I am just not understanding the documentation.

I welcome an explanation.

Thank you.

DG

Response:

So a couple thoughts. First is would there be any chance you have PC_POOLSIZE in your onconfig more then 1 time? If that is the case, it would take the last entry it found for the value. Second, it does appear that you could use onmode -wm to modify this parameter dynamically as long as you are increasing the size. You should be able to run onstat -g cfg full (will probably want to redirect the value to a file), then search the file for PC_POOLSIZE and it will show default, current, and onconfig values. So I would guess if the first idea isn't happening, then onconfig setting will be 127 and current I believe would show 566. That could happen via onmode -wm (so it was ran and only changed memory, not file). Lastly, a quick look at code does seem to show that for certain conditions while processing a cache misses, the server would increase the size of the cache by 10% of it's current size (whether this happens or does not appear to be controlled by AUTO_TUNE, it looks like the server has been coded to just increase the size to reduce certain types of cache misses...so not every cache miss would trigger the increase in size). I'm not 100% certain of this because I believe the server has 2 different implementations of caches and a quick look it wasn't easy to tell if the procedure cache is implemented in this manner or not. I would infer that it is using this method of caching if idea's 1 and 2 are not true, as then that would explain the increase in the cache size. A quick look at this code also didn't seem to show a way to limit growth in this case. Also, in this automatic growth case, the onstat -g cfg full would also show the current value as 566, so from just that output it wouldn't be possible to distinguish option 2 from option 3, as option 3 also ends up setting the "current" size displayed in the onstat -g cfg full.

Jacques Renaut
HCL Informix Advanced Support

Messages In This Thread

• PC_POOLSIZE?
  DAVID GROVE -- Wednesday, 21 August 2019, at 7:47 p.m.
  • Re: PC_POOLSIZE?
    JACQUES RENAUT -- Wednesday, 21 August 2019, at 9:15 p.m.
    • Re: PC_POOLSIZE?
      DAVID GROVE -- Thursday, 22 August 2019, at 2:25 p.m.
      • Re: PC_POOLSIZE?
        DAVID GROVE -- Thursday, 22 August 2019, at 2:42 p.m.
        • Re: PC_POOLSIZE?
          JACQUES RENAUT -- Thursday, 22 August 2019, at 3:06 p.m.

IDS Forum is maintained by Administrator with WebBBS 5.12.