shm error on Solaris 8
I saw a problem similar to mine on this thread: http://www.hswn.dk/hobbiton/2006/08/msg00667.html <http://www.hswn.dk/hobbiton/2006/08/msg00667.html> My situation may be a little different though. In my case I've been tasked with running a hobbit instance in parallel with an existing bb4 instance ON THE SAME HOST. I've modified my hobbitserver.cfg to run on port 1985 but I get the same error in /var/log/hobbitlaunch.log as David observed. A section of the logs follows (similar messages repeat): 2006-09-27 12:48:56 hobbitlaunch starting 2006-09-27 12:48:56 Loading tasklist configuration from /opt/home/hobbit/server/etc/hobbitlaunch.cfg 2006-09-27 12:48:56 Loading hostnames 2006-09-27 12:48:56 Loading saved state 2006-09-27 12:48:56 Cannot access checkpoint file /opt/home/hobbit/server/tmp/hobbitd.chk for restore 2006-09-27 12:48:56 Setting up network listener on 0.0.0.0:1985 2006-09-27 12:48:56 Setting up signal handlers 2006-09-27 12:48:56 Setting up hobbitd channels 2006-09-27 12:48:56 Could not attach shm Too many open files 2006-09-27 12:48:56 Cannot setup client channel 2006-09-27 12:48:56 Task hobbitd terminated, status 1
Help? I'm not comfortable tweaking kernel parameters. Can someone walk me through it? Should it solve my problems? Can hobbit and bb4 co-exist on the same host?
Stephen F. Menton メントン・スティーバン Cingular Wireless, Business Consultant I IT Billing Operations, Performance Monitoring stephen.menton at cingular.com (425) 288-7759 - office
I've modified /etc/system to include: set shmsys:shminfo_shmseg=7
and rebooted... but I still get the exact same error. What else must be changed for Solaris 8 ?
-----Original Message-----
From: Menton, Stephen
Sent: Wednesday, September 27, 2006 12:57 PM
To: hobbit at hswn.dk
Subject: [hobbit] shm error on Solaris 8
I saw a problem similar to mine on this thread: http://www.hswn.dk/hobbiton/2006/08/msg00667.html <http://www.hswn.dk/hobbiton/2006/08/msg00667.html>
My situation may be a little different though. In my case I've been tasked with running a hobbit instance in parallel with an existing bb4 instance ON THE SAME HOST. I've modified my hobbitserver.cfg to run on port 1985 but I get the same error in /var/log/hobbitlaunch.log as David observed. A section of the logs follows (similar messages repeat):
2006-09-27 12:48:56 hobbitlaunch starting
2006-09-27 12:48:56 Loading tasklist configuration from /opt/home/hobbit/server/etc/hobbitlaunch.cfg
2006-09-27 12:48:56 Loading hostnames
2006-09-27 12:48:56 Loading saved state
2006-09-27 12:48:56 Cannot access checkpoint file /opt/home/hobbit/server/tmp/hobbitd.chk for restore
2006-09-27 12:48:56 Setting up network listener on 0.0.0.0:1985
2006-09-27 12:48:56 Setting up signal handlers
2006-09-27 12:48:56 Setting up hobbitd channels
2006-09-27 12:48:56 Could not attach shm Too many open files
2006-09-27 12:48:56 Cannot setup client channel
2006-09-27 12:48:56 Task hobbitd terminated, status 1
Help? I'm not comfortable tweaking kernel parameters. Can someone walk me through it? Should it solve my problems? Can hobbit and bb4 co-exist on the same host?
Stephen F. Menton メントン・スティーバン
Cingular Wireless, Business Consultant I
IT Billing Operations, Performance Monitoring
stephen.menton at cingular.com
(425) 288-7759 - office
try running "ulimit -a" and see if you are being limited as to how many open files or memory you can access. If so you can change the limit to unlimited.
-Charles
Menton, Stephen wrote:
I've modified /etc/system to include: set shmsys:shminfo_shmseg=7 and rebooted... but I still get the exact same error. What else must be changed for Solaris 8 ?
-----Original Message----- *From:* Menton, Stephen *Sent:* Wednesday, September 27, 2006 12:57 PM *To:* hobbit at hswn.dk *Subject:* [hobbit] shm error on Solaris 8 I saw a problem similar to mine on this thread: http://www.hswn.dk/hobbiton/2006/08/msg00667.html My situation may be a little different though. In my case I've been tasked with running a hobbit instance in parallel with an existing bb4 instance ON THE SAME HOST. I've modified my hobbitserver.cfg to run on port 1985 but I get the same error in /var/log/hobbitlaunch.log as David observed. A section of the logs follows (similar messages repeat): 2006-09-27 12:48:56 hobbitlaunch starting 2006-09-27 12:48:56 Loading tasklist configuration from /opt/home/hobbit/server/etc/hobbitlaunch.cfg 2006-09-27 12:48:56 Loading hostnames 2006-09-27 12:48:56 Loading saved state 2006-09-27 12:48:56 Cannot access checkpoint file /opt/home/hobbit/server/tmp/hobbitd.chk for restore 2006-09-27 12:48:56 Setting up network listener on 0.0.0.0:1985 2006-09-27 12:48:56 Setting up signal handlers 2006-09-27 12:48:56 Setting up hobbitd channels 2006-09-27 12:48:56 Could not attach shm Too many open files 2006-09-27 12:48:56 Cannot setup client channel 2006-09-27 12:48:56 Task hobbitd terminated, status 1 Help? I'm not comfortable tweaking kernel parameters. Can someone walk me through it?Should it solve my problems? Can hobbit and bb4 co-exist on the same host? Stephen F. Menton メントン・スティーバン Cingular Wireless, Business Consultant I IT Billing Operations, Performance Monitoring stephen.menton at cingular.com <mailto:stephen.menton at cingular.com> (425) 288-7759 - office
Ahh, file descriptors is at it's default, 256. Does this need to be increased? I've tried doing a 'ulimit -n 1024' and confirmed the change before doing "./server/hobbit.sh start" and it hasn't changed the results.
Do I need to change something again in /etc/system to make this a permanent change ?
steve
-----Original Message-----
From: Charles Jones [mailto:jonescr at cisco.com]
Sent: Thursday, September 28, 2006 12:11 PM
To: hobbit at hswn.dk
Subject: Re: [hobbit] shm error on Solaris 8
try running "ulimit -a" and see if you are being limited as to how many open files or memory you can access. If so you can change the limit to unlimited.
-Charles
Menton, Stephen wrote:
I've modified /etc/system to include:
set shmsys:shminfo_shmseg=7
and rebooted... but I still get the exact same error. What else must be changed for Solaris 8 ?
-----Original Message-----
From: Menton, Stephen
Sent: Wednesday, September 27, 2006 12:57 PM
To: hobbit at hswn.dk
Subject: [hobbit] shm error on Solaris 8
I saw a problem similar to mine on this thread: http://www.hswn.dk/hobbiton/2006/08/msg00667.html <http://www.hswn.dk/hobbiton/2006/08/msg00667.html>
My situation may be a little different though. In my case I've been tasked with running a hobbit instance in parallel with an existing bb4 instance ON THE SAME HOST. I've modified my hobbitserver.cfg to run on port 1985 but I get the same error in /var/log/hobbitlaunch.log as David observed. A section of the logs follows (similar messages repeat):
2006-09-27 12:48:56 hobbitlaunch starting
2006-09-27 12:48:56 Loading tasklist configuration from /opt/home/hobbit/server/etc/hobbitlaunch.cfg
2006-09-27 12:48:56 Loading hostnames
2006-09-27 12:48:56 Loading saved state
2006-09-27 12:48:56 Cannot access checkpoint file /opt/home/hobbit/server/tmp/hobbitd.chk for restore
2006-09-27 12:48:56 Setting up network listener on 0.0.0.0:1985
2006-09-27 12:48:56 Setting up signal handlers
2006-09-27 12:48:56 Setting up hobbitd channels
2006-09-27 12:48:56 Could not attach shm Too many open files
2006-09-27 12:48:56 Cannot setup client channel
2006-09-27 12:48:56 Task hobbitd terminated, status 1
Help? I'm not comfortable tweaking kernel parameters. Can someone walk me through it? Should it solve my problems? Can hobbit and bb4 co-exist on the same host?
Stephen F. Menton メントン・スティーバン
Cingular Wireless, Business Consultant I
IT Billing Operations, Performance Monitoring
stephen.menton at cingular.com
(425) 288-7759 - office
Le 28 sept. 06 à 21:32, Menton, Stephen a écrit :
Ahh, file descriptors is at it's default, 256. Does this need to be
increased? I've tried doing a 'ulimit -n 1024' and confirmed the
change before doing "./server/hobbit.sh start" and it hasn't
changed the results.Do I need to change something again in /etc/system to make this a
permanent change ?
rlim_fd_cur
see http://docs.sun.com/app/docs/doc/816-0607/6m735r5fi?q=tunable +kernel&a=view
...but can't be that ( see previous mail )
Le 28 sept. 06 à 21:11, Charles Jones a écrit :
try running "ulimit -a" and see if you are being limited as to how
many open files or memory you can access. If so you can change the
limit to unlimited.
No. I don't think it could be the right way. (and this is limited to
1024...rlim_fd_max in /etc/system)
2006-09-27 12:48:56 Could not attach shm Too many open files
extract of man shmat :
EMFILE
The number of shared memory segments attached to the
calling process would exceed the system-imposed limit.
I have one box with 8 segment for hobbit.
Why don't you put shmseg to 32 and try again ? set shmsys:shminfo_shmseg=32 (I'd even set it to 64)
and also increase shminfo_shmmni set shmsys:shminfo_shmmni=512
http://docs.sun.com/app/docs/doc/816-0607/6m735r5fq?q=tunable +kernel&a=view
Nicolas
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=- Nicolas Dorfsman Tél: 06.7981.4486 Expert UNIX Fax: 05.4755.1045
mailto:ndo at unikservice.com http://www.unikservice.com
UNIKService est membre du groupement Opale Group (www.opalegroup.com)
Le 28 sept. 06 à 21:34, Nicolas Dorfsman a écrit :
Why don't you put shmseg to 32 and try again ? set shmsys:shminfo_shmseg=32 (I'd even set it to 64)
and also increase shminfo_shmmni set shmsys:shminfo_shmmni=512
Check also that your tunnables are really set :
sysdef -i | grep SHM
Nicolas
sysdef -i | grep SHM
1048576 max shared memory segment size (SHMMAX) 1 min shared memory segment size (SHMMIN) 100 shared memory identifiers (SHMMNI) 7 max attached shm segments per process (SHMSEG)
-----Original Message----- From: Nicolas Dorfsman [mailto:ndo at unikservice.com] Sent: Thursday, September 28, 2006 12:50 PM To: hobbit at hswn.dk Subject: Re: [hobbit] shm error on Solaris 8
Le 28 sept. 06 à 21:34, Nicolas Dorfsman a écrit :
Why don't you put shmseg to 32 and try again ? set shmsys:shminfo_shmseg=32 (I'd even set it to 64)
and also increase shminfo_shmmni set shmsys:shminfo_shmmni=512
Check also that your tunnables are really set :
sysdef -i | grep SHM
Nicolas To unsubscribe from the hobbit list, send an e-mail to hobbit-unsubscribe at hswn.dk
Le 28 sept. 06 à 21:54, Menton, Stephen a écrit :
sysdef -i | grep SHM
1048576 max shared memory segment size (SHMMAX) 1 min shared memory segment size (SHMMIN) 100 shared memory identifiers (SHMMNI) 7 max attached shm segments per process (SHMSEG)
wow
put that :
set shmsys:shminfo_shmmax=16777216 set shmsys:shminfo_shmseg=64 set shmsys:shminfo_shmmni=512
rlim_fd_max=2048 rlim_fd_cur=1024
participants (3)
-
jonescr@cisco.com
-
ndo@unikservice.com
-
stephen.menton@cingular.com