Re: shared memory problem

"Art S. Kagel" wrote: > > You kernel is not allowing the engine to create any more shared > memory segments. Build a new kernel increasing all relevant shared > memory parameters especially any that relate to the number of > segments a process is permitted. For now increase SHMVIRTSIZE and > SHMADD since it seems to be the number of segments that is the > problem not the size of one. > > Art S. Kagel > > Daniel Willems wrote: > > > > environment: > > server1 (ULLA): Compaq DS20E, 2x 500Mhz CPU, 1.5GB memory, BaaN IVc3, > > I-connenct 2.20, Unix True64 4.0F > > server2 (GELLA): Compaq DS20E, 2x 500Mhz CPU, 1.5GB memory, Informix > > 7.31.FC6, Unix True64 4.0F > > onconfig : SHMVIRTSIZE 524000, SHMADD 8192, SHMTOTAL 0 > > > > Problem: > > Informix crashes with > > 14:40:54 shmat: [ENOMEM][12]: out of available data space, check system> > MAXMEM > > 14:40:54 out of virtual shared memory > > 14:40:54 Assert Failed: No Exception Handler > > 14:40:54 Informix Dynamic Server Version 7.31.FC6 > > 14:40:54 Who: Session(572, kleinw@ULLA.soudal, 4623, 499379912) > > Thread(4108, sqlexec, 21dc2c4e8, 1) > > File: mtex.c Line: 446 > > 14:40:54 Results: Exception Caught. Type: MT_EX_BE, Context: mt_ex_setup:> > no mem > > 14:40:54 Action: Please notify Informix Technical Support. > > 14:41:15 See Also: /tmp/af.13f44766, shmem.13f44766.0 > > 14:41:22 Error writing '/tmp/shmem.13f44766.0' errno = 28 > > 14:41:22 mtex.c, line 446, thread 4108, proc id 880, No Exception Handler. > > 14:41:22 PANIC: Attempting to bring system down> > > > onstat -g seg shows a total size of 1015021568, details:> > class R* : size 142606336 > > class V : initial size 536870912, 39 x 8388608 added No, it is the total memory a single process can attach to. The default is 1 Gb. So increasing your initial segment size will prevent the server from booting at all. Also, you only have 1.5 Gb available. If you only have the database server on this machine, then that is okay. However, if you are actually running Baan on this machine, then you have a severe deficiency. Baan uses an inordinate amount of memory in it's own right. ******** Okay, I reread your post and I now see 2 servers. Excellent architecture, and not just for Baan. :-) If you still run short of memory, then reschedule your Baan jobs so that they don't overlap. I am assuming it is Baan batch processing of some sort that is causing this. Cheers, -- Mark. +----------------------------------------------------------+-----------+ | Mark D. Stock mailto:mdstock@mydas.freeserve.co.uk |//////// /| | http://www.informix.com http://www.informixhandbook.com |///// / //| | http://www.iiug.org +-----------------------------------+//// / ///| | |This email will self-destruct in |/// / ////| | |10 sec. If you received this email |// / /////| | |in error, sorry about the mess. |/ ////////| +----------------------+-----------------------------------+-----------+

To all, I changed SHMVIRTSIZE and SHMADD, but it was Unix that was limiting the shared memory to 1GB. That limit was removed and I'm now waiting till we hit the 1GB again. Thanks to all. Daniel Willems Mark D. Stock <mdstock@mydas.freeserve.co.uk> wrote in message news:8lq5e2$g0e$1@news.xmission.com... > > "Art S. Kagel" wrote: > > > > You kernel is not allowing the engine to create any more shared > > memory segments. Build a new kernel increasing all relevant shared > > memory parameters especially any that relate to the number of > > segments a process is permitted. For now increase SHMVIRTSIZE and > > SHMADD since it seems to be the number of segments that is the > > problem not the size of one. > > > > Art S. Kagel > > > > Daniel Willems wrote: > > > > > > environment: > > > server1 (ULLA): Compaq DS20E, 2x 500Mhz CPU, 1.5GB memory, BaaN IVc3, > > > I-connenct 2.20, Unix True64 4.0F > > > server2 (GELLA): Compaq DS20E, 2x 500Mhz CPU, 1.5GB memory, Informix > > > 7.31.FC6, Unix True64 4.0F > > > onconfig : SHMVIRTSIZE 524000, SHMADD 8192, SHMTOTAL 0 > > > > > > Problem: > > > Informix crashes with > > > 14:40:54 shmat: [ENOMEM][12]: out of available data space, checksystem > > > MAXMEM > > > 14:40:54 out of virtual shared memory > > > 14:40:54 Assert Failed: No Exception Handler > > > 14:40:54 Informix Dynamic Server Version 7.31.FC6 > > > 14:40:54 Who: Session(572, kleinw@ULLA.soudal, 4623, 499379912) > > > Thread(4108, sqlexec, 21dc2c4e8, 1) > > > File: mtex.c Line: 446 > > > 14:40:54 Results: Exception Caught. Type: MT_EX_BE, Context: mt_ex_setup:> > > no mem > > > 14:40:54 Action: Please notify Informix Technical Support. > > > 14:41:15 See Also: /tmp/af.13f44766, shmem.13f44766.0 > > > 14:41:22 Error writing '/tmp/shmem.13f44766.0' errno = 28 > > > 14:41:22 mtex.c, line 446, thread 4108, proc id 880, No Exception Handler. > > > 14:41:22 PANIC: Attempting to bring system down> > > > > > onstat -g seg shows a total size of 1015021568, details:> > > class R* : size 142606336 > > > class V : initial size 536870912, 39 x 8388608 added > > No, it is the total memory a single process can attach to. The default > is 1 Gb. So increasing your initial segment size will prevent the server > from booting at all. Also, you only have 1.5 Gb available. If you only > have the database server on this machine, then that is okay. However, if > you are actually running Baan on this machine, then you have a severe > deficiency. Baan uses an inordinate amount of memory in it's own right. > > ******** Okay, I reread your post and I now see 2 servers. Excellent > architecture, and not just for Baan. :-) > > If you still run short of memory, then reschedule your Baan jobs so that > they don't overlap. I am assuming it is Baan batch processing of some > sort that is causing this. > > Cheers, > -- > Mark. > > +----------------------------------------------------------+-----------+ > | Mark D. Stock mailto:mdstock@mydas.freeserve.co.uk |//////// /| > | http://www.informix.com http://www.informixhandbook.com |///// / file://| > | http://www.iiug.org +-----------------------------------+//// / ///| > | |This email will self-destruct in |/// / ////| > | |10 sec. If you received this email |// / /////| > | |in error, sorry about the mess. |/ ////////| > +----------------------+-----------------------------------+-----------+