Re: Informix RESIDENT and Solaris 8 and 10 swap usage

On Jun 29, 10:17 am, Koya <chav...@gmail.com> wrote: > On Jun 29, 8:42 am, Eric Rowell <erow...@gmail.com> wrote: > > > > > > > Koya , > >     Wish I still had access to my old Solaris systems (past live), this is > > what I remember. > > >     There was a change to the mapping and controls of Solaris shared > > memory.   Solaris started tracking shared memory usage which when > > running Inforix or other applications which share memory caused memory usage > > to sky rocket.  There is a mapping page for the shared memory which is used > > by the OS to protect memory and track the application which currently has > > access to the mapped page.   I think we had to set something to stop this > > from happening (Solaris setting).  Once we did memory worked as it did in > > the past and it was up to the application to protect it's self.   I cannot > > find the details in my notes since the SA's handled it.  I hope this helps > > get you in the right direction. > > > Eric > > > On Fri, Jun 26, 2009 at 3:55 PM, Koya <chav...@gmail.com> wrote: > > > Hello all, > > > > I  have the follwoing question about RESIDENT and swap usage in > > > Solaris. > > > > My understanding is if we set RESIDENT value greater than zero, swap > > > should not be used. If this is corrent, do you have any idea why swap > > > is in use ? > > > > top command in Solaris 8 displaying 29.9G swap in use, but we have > > > RESIDENT -1 ( which enables memory to be resident, we have total 3 > > > instances on this server using 29 GB of memory, the following onstat > > > output shows 27GB on one instance, other two instances use 1.8 GB) > > > > onstat - > > > IBM Informix Dynamic Server Version 10.00.FC10    -- On-Line -- Up 27 > > > days 03:41:08 -- 27467776 Kbytes> > > > both in Sol 8 and 10 "swap" command displaying swap used ( 31GB and > > > 33GB). > > > > swap -s > > > total: 29885536k bytes allocated + 1466440k reserved = 31351976k used, > > > 17550648k available > > > > top command output ( Solaris 8 ) : > > > > load averages:  3.59,  2.56,  2.21         14:14:49 > > > 220 processes: 213 sleeping, 1 zombie, 3 stopped, 3 on cpu > > > CPU states: 78.1% idle, 13.0% user,  3.6% kernel,  5.2% iowait,  0.0% > > > swap > > > Memory: 48.0G real, 13.6G free, 29.9G swap in use, 16.7G swap free > > > >   PID USERNAME THR PR NCE  SIZE   RES STATE   TIME FLTS    CPU > > > COMMAND > > >  23520 informix   5 30 -10 26.2G 26.2G cpu23 125.7H    0  3.04% oninit > > >  23526 informix   5 59 -10 26.2G 26.2G sleep 174.5H    0  2.55% oninit > > >  23506 informix   5 59 -10 26.2G 26.2G sleep 132.2H    0  2.24% oninit > > >  23521 informix   5 59 -10 26.2G 26.2G sleep 110.9H    0  1.88% oninit > > >  23522 informix   5 59 -10 26.2G 26.2G sleep  98.4H    0  1.84% oninit > > >  23525 informix   5 59 -10 26.2G 26.2G sleep 189.7H    0  1.57% oninit > > >  23527 informix   5 50 -10 26.2G 26.2G sleep 163.9H    0  1.34% oninit > > >  23524 informix   5 59 -10 26.2G 26.2G sleep  77.5H    0  1.17% oninit > > >  23529 informix   5 59 -10 26.2G 26.2G sleep 151.5H    0  1.01% oninit > > >  23523 informix   5 59 -10 26.2G 26.2G sleep  88.5H    0  1.00% oninit > > >  23528 informix   5 59 -10 26.2G 26.2G sleep 156.6H    0  0.88% oninit > > > > Solaris 10 : > > > ======== > > > > We have Solaris 10 server has total 5 instances assigned/using 23 GB, > > > top command display is little bit different, which does not show "swap > > > in use ", but total swap ( does this mean swap in use ? think so > > > because available is another 47GB) > > > > swap -s > > > total: 16559736k bytes allocated + 16692216k reserved = 33251952k > > > used, 47252080k available > > > > load averages:  1.90,  1.91,  1.98;                    up > > > 12+12:04:20 > > > 14:36:17 > > > 223 processes: 214 sleeping, 1 zombie, 2 stopped, 6 on cpu > > > CPU states: 87.8% idle,  7.2% user,  5.0% kernel,  0.0% iowait,  0.0% > > > swap > > > Memory: 48G phys mem, 24G free mem, 38G total swap, 38G free swap > > > >   PID USERNAME LWP PRI NICE  SIZE   RES STATE    TIME    CPU COMMAND > > >  5622 root       1  50    0   10M 7368K cpu/17   0:08  4.13% cputst > > >  1261 informix   1  59  -10 2887M  879M cpu/3   94:19  1.80% oninit > > >  20364 root      11  59    0   26M   15M sleep   58:17  0.78% bpbkar > > >   279 informix   2  59  -10 7706M 2984M cpu/16 349:04  0.48% oninit > > >   283 informix   2  59  -10 7706M 2977M cpu/9  308:43  0.43% oninit > > >   170 informix   2  59  -10 7706M 2997M sleep  286:40  0.41% oninit > > >   281 informix   2  59  -10 7706M 2981M sleep  310:20  0.41% oninit > > >   280 informix   2  59  -10 7706M 2983M sleep  329:45  0.40% oninit > > >   282 informix   2  59  -10 7706M 2973M cpu/19 260:14  0.39% oninit > > >   239 informix   2  59  -10 7706M 2991M sleep  334:40  0.35% oninit > > >  1257 informix   2  59  -10 2889M 1068M sleep  204:39  0.25% oninit > > > > -- Chavan > > > _______________________________________________ > > > Informix-list mailing list > > > Informix-l...@iiug.org > > >http://www.iiug.org/mailman/listinfo/informix-list > > > -- > > Eric B. Rowell- Hide quoted text - > > > - Show quoted text - > > Hi Eric, >            Thanks for the reply. I am looking into Solaris Internals. > Will let you know if I find someting. > -- Chavan- Hide quoted text - > > - Show quoted text - I got the follwoing info from unix admin. Which explains why. Swap –l shows you the actual swap space configured on the disk, in which case the following output shows: root# swap -l swapfile dev swaplo blocks free /dev/md/dsk/d31 85,31 16 16779296 16764688 This means that all of the swap space pages on the disk is currently not in use so all of the mem pages for the database is actually in physical memory. swap –l only lists the swap areas and does not include swap space in the form of physical memory. So top command reports the swap space in the form of physical memory. -- Chavan