2 shared memory segments

Informix 10.00.HC5 HP-UX 11i v2 Recently, I've started getting message like the following when I restart the server Could not create single shared memory segment with resident and non-resident partitions. Proceeding to create 2 shared memory segments instead. Sometimes, the server startup continues OK but other times the above message is followed by the following or similar message: shmget: [ENOMEM][12]: key 52574802: out of shared memory, check system max shared memory segment size mt_shm_init: can't create virtual segment I'm guessing it's maybe a kernel parameter (shmmax?) issue but I haven't hit on the correct one to increase or how much to increase it by. shmmax = 307200000000 SHMVIRTSIZE = 321536 Buffers = 350000 (2k) Any ideas what to tune and what to tune it to? Thanks! Brian

Hi Brian, In addition it'd be worthwhile seeing if anything else is using shared memory. Again ipcs should help you here. During the flurry of activity at startup it may be that Informix and something else are both attempting to allocate memory and depending on who gets in first will determine whether you get the error or not. Best wishes, Andy. > To: ids@iiug.org > From: brucemarchu@aol.com > Subject: Re: 2 shared memory segments [18848 [18849] > Date: Fri, 29 Jan 2010 20:36:14 -0500 > > Hi Brian, > > Here are several things to consider in addition to onconfig parameters: > 1. There might be a problem with not enough available memory due to memor= > y useage by applications that are running on the server. When you say "re= > start the server" do you mean just restarting the Informix engine, or are= > you rebooting the actual physical machine? If your UNIX machine has acce= > ss to the "top" command you can run top with various flags to see what pro= > cesses are using the most cpu's and the related memory being used. > 2. Is the Informix engine crashing or are you doing an orderly shutdown= > and then a restart? If the Informix engine is hanging or crashing, then= > are you checking to make sure that all of the Informix memory segments ha= > ve been cleaned out before a restart. You can use ipcs and various flags= > such as ipcs -a, and ipcs -m from the UNIX prompt. > 3. Check the Informix machine and documentation notes to see settings for= > various UNIX kernel parameters relating to memory, and recheck your curre= > nt settings. Make sure that you have at least the minimum values shown fo= > r memory related settings. Note that changing certain kernel parameters mi= > ght require a reboot of the physical machine. =20 > 4. Recheck the setting for the starting memory location SHMBASE in the onc= > onfig file. > 5. Recheck the UNIX system logs [ maybe at /var/adm/syslog or a similar= > location] to see if there are any error message regarding memory. > > Regards, > Bruce > > -----Original Message----- > From: Brian Amstutz <brian.amstutz@asburyseminary.edu> > To: ids@iiug.org > Sent: Fri, Jan 29, 2010 12:31 pm > Subject: 2 shared memory segments [18848] > > Informix 10.00.HC5 HP-UX 11i v2=20 > Recently, I've started getting message like the following when I restart= > =20 > he server=20 > Could not create single shared memory segment with resident and=20 > on-resident partitions. Proceeding to create 2 shared memory segments=20 > nstead.=20 > Sometimes, the server startup continues OK but other times the above=20 > essage is followed by the following or similar message:=20 > shmget: [ENOMEM][12]: key 52574802: out of shared memory, check system max= > =20 > hared memory segment size=20 > t_shm_init: can't create virtual segment=20 > I'm guessing it's maybe a kernel parameter (shmmax?) issue but I haven't= > =20 > it on the correct one to increase or how much to increase it by.=20 > shmmax =3D 307200000000=20 > HMVIRTSIZE =3D 321536=20 > uffers =3D 350000 (2k)=20 > Any ideas what to tune and what to tune it to?=20 > Thanks!=20 > Brian=20 > > **************************************************************************= > ****=20 > Forum Note: Use "Reply" to post a response in the discussion forum.=20 > > > ******************************************************************************* > Forum Note: Use "Reply" to post a response in the discussion forum. > _________________________________________________________________ We want to hear all your funny, exciting and crazy Hotmail stories. Tell us now http://clk.atdmt.com/UKM/go/195013117/direct/01/

Thanks Bruce and Andy. I have received this message both when only restarting Informix AND when rebooting the entire server. I've used ipcs to try to find out what else is using shared memory. I've identified a couple of apps and placed Informix ahead of those in the startup order. However, there are some ipcs entries whose CPID or LPID do not show up in the current list of processes - not sure if this is "normal" under HPUX or not??? 'top' nearly always shows the 3 informix cpu vp's as the top 3 process - at least all three within the top 8 processes. I checked the machine notes and the only kernel parameter that wasn't set high enough was semmnu - I bumped it up from 256 to 2048 and will have to a find time to reboot the system. The recommendation for shmmax is anywhere from 4Gb up to the maximum allowed. Currently, ours is set to around 286Gb and the max allowed is 4096Gb. I may not understand exactly how shmmax is used but with only a 22Gb database, 8Gb RAM and 300Gb total disk space - it seems like 286Gb for the max shared memory size should be plenty large; however, I have bumped this up to approx 500Gb Recommended SHMBASE is 0x0L - ours is set to 0x0 (minus the L) which I *THINK* is the same thing??? Nothing of interest found in the syslogs Brian >-----Original Message----- >From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of >brucemarchu@aol.com >Sent: Friday, January 29, 2010 8:36 PM >To: ids@iiug.org >Subject: Re: 2 shared memory segments [18848 [18849] > >Hi Brian, > >Here are several things to consider in addition to onconfig parameters: >1. There might be a problem with not enough available memory due to memor= >y useage by applications that are running on the server. When you say "re= >start the server" do you mean just restarting the Informix engine, or are= >you rebooting the actual physical machine? If your UNIX machine has acce= >ss to the "top" command you can run top with various flags to see what pro= >cesses are using the most cpu's and the related memory being used. >2. Is the Informix engine crashing or are you doing an orderly shutdown= >and then a restart? If the Informix engine is hanging or crashing, then= >are you checking to make sure that all of the Informix memory segments ha= >ve been cleaned out before a restart. You can use ipcs and various flags= >such as ipcs -a, and ipcs -m from the UNIX prompt. >3. Check the Informix machine and documentation notes to see settings for= >various UNIX kernel parameters relating to memory, and recheck your curre= >nt settings. Make sure that you have at least the minimum values shown fo= >r memory related settings. Note that changing certain kernel parameters mi= >ght require a reboot of the physical machine. =20 >4. Recheck the setting for the starting memory location SHMBASE in the onc= >onfig file. >5. Recheck the UNIX system logs [ maybe at /var/adm/syslog or a similar= >location] to see if there are any error message regarding memory. > >Regards, >Bruce > >-----Original Message----- >From: Brian Amstutz <brian.amstutz@asburyseminary.edu> >To: ids@iiug.org >Sent: Fri, Jan 29, 2010 12:31 pm >Subject: 2 shared memory segments [18848] > >Informix 10.00.HC5 HP-UX 11i v2=20 >Recently, I've started getting message like the following when I restart= >=20 >he server=20 >Could not create single shared memory segment with resident and=20 >on-resident partitions. Proceeding to create 2 shared memory segments=20 >nstead.=20 >Sometimes, the server startup continues OK but other times the above=20 >essage is followed by the following or similar message:=20 >shmget: [ENOMEM][12]: key 52574802: out of shared memory, check system max= >=20 >hared memory segment size=20 >t_shm_init: can't create virtual segment=20 >I'm guessing it's maybe a kernel parameter (shmmax?) issue but I haven't= >=20 >it on the correct one to increase or how much to increase it by.=20 >shmmax =3D 307200000000=20 >HMVIRTSIZE =3D 321536=20 >uffers =3D 350000 (2k)=20 >Any ideas what to tune and what to tune it to?=20 >Thanks!=20 >Brian=20 > >************************************************************************* *= >****=20 >Forum Note: Use "Reply" to post a response in the discussion forum.=20 > > >************************************************************************* ****** > Forum Note: Use "Reply" to post a response in the discussion forum.

Hi Brian, I have seen the SHMBASE parameter used with and without the "L". What= do you see when you run the "onstat -g seg" command? Since you also have the problem when rebooting the server you could consid= er changing the order in which processes and applications are started on= the computer after a reboot to have the Informix engine start up before= a lot of other applications and processes. Take a look at the related st= art up scripts such at the iniit.d and rc scripts. See what is being brou= ght up before the Informix. Regards, Bruce -----Original Message----- From: Brian Amstutz <brian.amstutz@asburyseminary.edu> To: ids@iiug.org Sent: Mon, Feb 1, 2010 1:06 pm Subject: RE: 2 shared memory segments [18848 [18849] [18862] Thanks Bruce and Andy.=20 I have received this message both when only restarting Informix AND when= =20 ebooting the entire server. I've used ipcs to try to find out what else=20 s using shared memory. I've identified a couple of apps and placed=20 nformix ahead of those in the startup order. However, there are some=20 pcs entries whose CPID or LPID do not show up in the current list of=20 rocesses - not sure if this is "normal" under HPUX or not???=20 'top' nearly always shows the 3 informix cpu vp's as the top 3 process -= =20 t least all three within the top 8 processes.=20 I checked the machine notes and the only kernel parameter that wasn't set= =20 igh enough was semmnu - I bumped it up from 256 to 2048 and will have to= =20 find time to reboot the system. The recommendation for shmmax is=20 nywhere from 4Gb up to the maximum allowed. Currently, ours is set to=20 round 286Gb and the max allowed is 4096Gb. I may not understand exactly=20 ow shmmax is used but with only a 22Gb database, 8Gb RAM and 300Gb total= =20 isk space - it seems like 286Gb for the max shared memory size should be= =20 lenty large; however, I have bumped this up to approx 500Gb=20 Recommended SHMBASE is 0x0L - ours is set to 0x0 (minus the L) which I=20 THINK* is the same thing???=20 Nothing of interest found in the syslogs=20 Brian=20 >-----Original Message-----=20 From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of=20 brucemarchu@aol.com=20 Sent: Friday, January 29, 2010 8:36 PM=20 To: ids@iiug.org=20 Subject: Re: 2 shared memory segments [18848 [18849]=20 =20 Hi Brian,=20 =20 Here are several things to consider in addition to onconfig parameters:=20 1. There might be a problem with not enough available memory due to=20 emor=3D=20 y useage by applications that are running on the server. When you say=20 re=3D=20 start the server" do you mean just restarting the Informix engine, or=20 re=3D=20 you rebooting the actual physical machine? If your UNIX machine has acce= =3D=20 ss to the "top" command you can run top with various flags to see what=20 ro=3D=20 cesses are using the most cpu's and the related memory being used.=20 2. Is the Informix engine crashing or are you doing an orderly shutdown=3D= =20 and then a restart? If the Informix engine is hanging or crashing, then=3D= =20 are you checking to make sure that all of the Informix memory segments=20 a=3D=20 ve been cleaned out before a restart. You can use ipcs and various flags= =3D=20 such as ipcs -a, and ipcs -m from the UNIX prompt.=20 3. Check the Informix machine and documentation notes to see settings=20 or=3D=20 various UNIX kernel parameters relating to memory, and recheck your=20 urre=3D=20 nt settings. Make sure that you have at least the minimum values shown=20 o=3D=20 r memory related settings. Note that changing certain kernel parameters=20 i=3D=20 ght require a reboot of the physical machine. =3D20=20 4. Recheck the setting for the starting memory location SHMBASE in the=20 nc=3D=20 onfig file.=20 5. Recheck the UNIX system logs [ maybe at /var/adm/syslog or a similar=3D= =20 location] to see if there are any error message regarding memory.=20 =20 Regards,=20 Bruce=20 =20 -----Original Message-----=20 From: Brian Amstutz <brian.amstutz@asburyseminary.edu>=20 To: ids@iiug.org=20 Sent: Fri, Jan 29, 2010 12:31 pm=20 Subject: 2 shared memory segments [18848]=20 =20 Informix 10.00.HC5 HP-UX 11i v2=3D20=20 Recently, I've started getting message like the following when I restart= =3D=20 =3D20=20 he server=3D20=20 Could not create single shared memory segment with resident and=3D20=20 on-resident partitions. Proceeding to create 2 shared memory segments=3D20= =20 nstead.=3D20=20 Sometimes, the server startup continues OK but other times the above=3D20= =20 essage is followed by the following or similar message:=3D20=20 shmget: [ENOMEM][12]: key 52574802: out of shared memory, check system=20 ax=3D=20 =3D20=20 hared memory segment size=3D20=20 t_shm_init: can't create virtual segment=3D20=20 I'm guessing it's maybe a kernel parameter (shmmax?) issue but I haven't= =3D=20 =3D20=20 it on the correct one to increase or how much to increase it by.=3D20=20 shmmax =3D3D 307200000000=3D20=20 HMVIRTSIZE =3D3D 321536=3D20=20 uffers =3D3D 350000 (2k)=3D20=20 Any ideas what to tune and what to tune it to?=3D20=20 Thanks!=3D20=20 Brian=3D20=20 =20 *************************************************************************= =20 =3D=20 ****=3D20=20 Forum Note: Use "Reply" to post a response in the discussion forum.=3D20= =20 =20 =20 *************************************************************************= =20 *****=20 Forum Note: Use "Reply" to post a response in the discussion forum.=20 **************************************************************************= ****=20 Forum Note: Use "Reply" to post a response in the discussion forum.=20

Thanks Bruce Posting my onstat -g seg looked like it was going to be pretty messy with line wrapping and all - I'm not exactly sure what I should be looking for in regards to SHMBASE. The addr of my (R)esident segment is 80000000. Is that what you were referring to? I have moved some things around in the startup process - I have to replace some hardware Friday so we'll see what happens went it reboots. Brian >-----Original Message----- >From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of >brucemarchu@aol.com >Sent: Tuesday, February 02, 2010 10:40 PM >To: ids@iiug.org >Subject: Re: 2 shared memory segments [18848 [18849] [18862 [18875] > >Hi Brian, > >I have seen the SHMBASE parameter used with and without the "L". What= >do you see when you run the "onstat -g seg" command? > >Since you also have the problem when rebooting the server you could consid= >er changing the order in which processes and applications are started on= >the computer after a reboot to have the Informix engine start up before= >a lot of other applications and processes. Take a look at the related st= >art up scripts such at the iniit.d and rc scripts. See what is being brou= >ght up before the Informix. > >Regards, >Bruce > >-----Original Message----- >From: Brian Amstutz <brian.amstutz@asburyseminary.edu> >To: ids@iiug.org >Sent: Mon, Feb 1, 2010 1:06 pm >Subject: RE: 2 shared memory segments [18848 [18849] [18862] > >Thanks Bruce and Andy.=20 >I have received this message both when only restarting Informix AND when= >=20 >ebooting the entire server. I've used ipcs to try to find out what else=20 >s using shared memory. I've identified a couple of apps and placed=20 >nformix ahead of those in the startup order. However, there are some=20 >pcs entries whose CPID or LPID do not show up in the current list of=20 >rocesses - not sure if this is "normal" under HPUX or not???=20 >'top' nearly always shows the 3 informix cpu vp's as the top 3 process -= >=20 >t least all three within the top 8 processes.=20 >I checked the machine notes and the only kernel parameter that wasn't set= >=20 >igh enough was semmnu - I bumped it up from 256 to 2048 and will have to= >=20 >find time to reboot the system. The recommendation for shmmax is=20 >nywhere from 4Gb up to the maximum allowed. Currently, ours is set to=20 >round 286Gb and the max allowed is 4096Gb. I may not understand exactly=20 >ow shmmax is used but with only a 22Gb database, 8Gb RAM and 300Gb total= >=20 >isk space - it seems like 286Gb for the max shared memory size should be= >=20 >lenty large; however, I have bumped this up to approx 500Gb=20 >Recommended SHMBASE is 0x0L - ours is set to 0x0 (minus the L) which I=20 >THINK* is the same thing???=20 >Nothing of interest found in the syslogs=20 >Brian=20 >>-----Original Message-----=20 >From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of=20 >brucemarchu@aol.com=20 >Sent: Friday, January 29, 2010 8:36 PM=20 >To: ids@iiug.org=20 >Subject: Re: 2 shared memory segments [18848 [18849]=20 >=20 >Hi Brian,=20 >=20 >Here are several things to consider in addition to onconfig parameters:=20 >1. There might be a problem with not enough available memory due to=20 >emor=3D=20 >y useage by applications that are running on the server. When you say=20 >re=3D=20 >start the server" do you mean just restarting the Informix engine, or=20 >re=3D=20 >you rebooting the actual physical machine? If your UNIX machine has acce= >=3D=20 >ss to the "top" command you can run top with various flags to see what=20 >ro=3D=20 >cesses are using the most cpu's and the related memory being used.=20 >2. Is the Informix engine crashing or are you doing an orderly shutdown=3D= >=20 >and then a restart? If the Informix engine is hanging or crashing, then=3D= >=20 >are you checking to make sure that all of the Informix memory segments=20 >a=3D=20 >ve been cleaned out before a restart. You can use ipcs and various flags= >=3D=20 >such as ipcs -a, and ipcs -m from the UNIX prompt.=20 >3. Check the Informix machine and documentation notes to see settings=20 >or=3D=20 >various UNIX kernel parameters relating to memory, and recheck your=20 >urre=3D=20 >nt settings. Make sure that you have at least the minimum values shown=20 >o=3D=20 >r memory related settings. Note that changing certain kernel parameters=20 >i=3D=20 >ght require a reboot of the physical machine. =3D20=20 >4. Recheck the setting for the starting memory location SHMBASE in the=20 >nc=3D=20 >onfig file.=20 >5. Recheck the UNIX system logs [ maybe at /var/adm/syslog or a similar=3D= >=20 >location] to see if there are any error message regarding memory.=20 >=20 >Regards,=20 >Bruce=20 >=20 >-----Original Message-----=20 >From: Brian Amstutz <brian.amstutz@asburyseminary.edu>=20 >To: ids@iiug.org=20 >Sent: Fri, Jan 29, 2010 12:31 pm=20 >Subject: 2 shared memory segments [18848]=20 >=20 >Informix 10.00.HC5 HP-UX 11i v2=3D20=20 >Recently, I've started getting message like the following when I restart= >=3D=20 >=3D20=20 >he server=3D20=20 >Could not create single shared memory segment with resident and=3D20=20 >on-resident partitions. Proceeding to create 2 shared memory segments=3D20= >=20 >nstead.=3D20=20 >Sometimes, the server startup continues OK but other times the above=3D20= >=20 >essage is followed by the following or similar message:=3D20=20 >shmget: [ENOMEM][12]: key 52574802: out of shared memory, check system=20 >ax=3D=20 >=3D20=20 >hared memory segment size=3D20=20 >t_shm_init: can't create virtual segment=3D20=20 >I'm guessing it's maybe a kernel parameter (shmmax?) issue but I haven't= >=3D=20 >=3D20=20 >it on the correct one to increase or how much to increase it by.=3D20=20 >shmmax =3D3D 307200000000=3D20=20 >HMVIRTSIZE =3D3D 321536=3D20=20 >uffers =3D3D 350000 (2k)=3D20=20 >Any ideas what to tune and what to tune it to?=3D20=20 >Thanks!=3D20=20 >Brian=3D20=20 >=20 >************************************************************************* = >=20 >=3D=20 >****=3D20=20 >Forum Note: Use "Reply" to post a response in the discussion forum.=3D20= >=20 >=20 >=20 >*****

After rebooting the server (with BUFFERPOOL set to 500000), I received the "2 shared memory segments" message but the database started up OK. However, a second instance failed to start with an "out of shared memory" error. I set BUFFERPOOL down to 400000 and rebooted the (entire) server again. Same messages as above; also, after it was up just a few minutes end users couldn't use ssh to login (syslog shows "mmap Not enough space" errors). So, I shutdown Informix (only) set BUFFERPOOL back to 350000 (where I started), shutdown a couple of other applications that were using shared memory, removed a couple of other shared memory segments that didn't have processes attached to them (but couldn't get rid of 3 others) and then restarted Informix. It came up with the "2 shared memory segments" but the second instance DID startup OK and users are able to log in. So, basically I'm back where I started. I'm still a little confused as to why there's not enough shared memory. Top shows more 1 Gb - 3 Gb free. Is there a (HPUX) command to show how much shared memory space is still available? Informix 10.00.HC5 HP-UX 11i v2 Brian >-----Original Message----- >From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of Brian Amstutz >Sent: Wednesday, February 03, 2010 10:06 AM >To: ids@iiug.org >Subject: RE: 2 shared memory segments [18848 [18849] [1.... [18881] > >Thanks Bruce > >Posting my onstat -g seg looked like it was going to be pretty messy with >line wrapping and all - I'm not exactly sure what I should be looking for >in regards to SHMBASE. The addr of my (R)esident segment is 80000000. Is >that what you were referring to? > >I have moved some things around in the startup process - I have to replace >some hardware Friday so we'll see what happens went it reboots. > >Brian > >>-----Original Message----- > >>From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of > >>brucemarchu@aol.com > >>Sent: Tuesday, February 02, 2010 10:40 PM > >>To: ids@iiug.org > >>Subject: Re: 2 shared memory segments [18848 [18849] [18862 [18875] > >> > >>Hi Brian, > >> > >>I have seen the SHMBASE parameter used with and without the "L". What= > >>do you see when you run the "onstat -g seg" command? > >> > >>Since you also have the problem when rebooting the server you could >consid= > >>er changing the order in which processes and applications are started on= > >>the computer after a reboot to have the Informix engine start up before= > >>a lot of other applications and processes. Take a look at the related st= > >>art up scripts such at the iniit.d and rc scripts. See what is being >brou= > >>ght up before the Informix. > >> > >>Regards, > >>Bruce > >> > >>-----Original Message----- > >>From: Brian Amstutz <brian.amstutz@asburyseminary.edu> > >>To: ids@iiug.org > >>Sent: Mon, Feb 1, 2010 1:06 pm > >>Subject: RE: 2 shared memory segments [18848 [18849] [18862] > >> > >>Thanks Bruce and Andy.=20 > >>I have received this message both when only restarting Informix AND when= > >>=20 > >>ebooting the entire server. I've used ipcs to try to find out what >else=20 > >>s using shared memory. I've identified a couple of apps and placed=20 > >>nformix ahead of those in the startup order. However, there are some=20 > >>pcs entries whose CPID or LPID do not show up in the current list of=20 > >>rocesses - not sure if this is "normal" under HPUX or not???=20 > >>'top' nearly always shows the 3 informix cpu vp's as the top 3 process -= > >>=20 > >>t least all three within the top 8 processes.=20 > >>I checked the machine notes and the only kernel parameter that wasn't >set= > >>=20 > >>igh enough was semmnu - I bumped it up from 256 to 2048 and will have to= > >>=20 > >>find time to reboot the system. The recommendation for shmmax is=20 > >>nywhere from 4Gb up to the maximum allowed. Currently, ours is set to=20 > >>round 286Gb and the max allowed is 4096Gb. I may not understand >exactly=20 > >>ow shmmax is used but with only a 22Gb database, 8Gb RAM and 300Gb total= > >>=20 > >>isk space - it seems like 286Gb for the max shared memory size should be= > >>=20 > >>lenty large; however, I have bumped this up to approx 500Gb=20 > >>Recommended SHMBASE is 0x0L - ours is set to 0x0 (minus the L) which I=20 > >>THINK* is the same thing???=20 > >>Nothing of interest found in the syslogs=20 > >>Brian=20 > >>>-----Original Message-----=20 > >>From: ids-bounces@iiug.org [mailto:ids-bounces@iiug.org] On Behalf Of=20 > >>brucemarchu@aol.com=20 > >>Sent: Friday, January 29, 2010 8:36 PM=20 > >>To: ids@iiug.org=20 > >>Subject: Re: 2 shared memory segments [18848 [18849]=20 > >>=20 > >>Hi Brian,=20 > >>=20 > >>Here are several things to consider in addition to onconfig >parameters:=20 > >>1. There might be a problem with not enough available memory due to=20 > >>emor=3D=20 > >>y useage by applications that are running on the server. When you say=20 > >>re=3D=20 > >>start the server" do you mean just restarting the Informix engine, or=20 > >>re=3D=20 > >>you rebooting the actual physical machine? If your UNIX machine has acce= > >>=3D=20 > >>ss to the "top" command you can run top with various flags to see what=20 > >>ro=3D=20 > >>cesses are using the most cpu's and the related memory being used.=20 > >>2. Is the Informix engine crashing or are you doing an orderly >shutdown=3D= > >>=20 > >>and then a restart? If the Informix engine is hanging or crashing, >then=3D= > >>=20 > >>are you checking to make sure that all of the Informix memory segments=20 > >>a=3D=20 > >>ve been cleaned out before a restart. You can use ipcs and various flags= > >>=3D=20 > >>such as ipcs -a, and ipcs -m from the UNIX prompt.=20 > >>3. Check the Informix machine and documentation notes to see settings=20 > >>or=3D=20 > >>various UNIX kernel parameters relating to memory, and recheck your=20 > >>urre=3D=20 > >>nt settings. Make sure that you have at least the minimum values shown=20 > >>o=3D=20 > >>r memory related settings. Note that changing certain kernel >parameters=20 > >>i=3D=20 > >>ght require a reboot of the physical machine. =3D20=20 > >>4. Recheck the setting for the starting memory location SHMBASE in the=20 > >>nc=3D=20 > >>onfig file.=20 > >>5. Recheck the UNIX system logs [ maybe at /var/adm/syslog or a >similar=3D= > >>=20 > >>location] to see if there are any error message regarding memory.=20 > >>=20 > >>Regards,=20 > >>Bruce=20 > >>=20 > >>