Re: [osol-discuss] svc.configd leaking on snv_129?
* Chavdar Ivanov [2009-12-15 15:04]: > Yesterday I upgraded three of my OpenSolaris systems to 129; this afternoon I > came to find one of them very sluggish; it turned out that: > > PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP > 11 root 1536M 1048M sleep 590 1:09:27 6.3% svc.configd/21 > 4034 postgres 154M 135M sleep 590 0:00:01 0.0% postgres/1 > 3917 postgres 154M 135M sleep 590 0:00:00 0.0% postgres/1 > 4027 postgres 154M 135M sleep 590 0:00:07 0.0% postgres/1 > 4028 postgres 154M 134M sleep 590 0:00:05 0.0% postgres/1 > (the rest were less significant). > > I disabled postgres for the moment and got the system to almost normal > responsiveness. Don't know what to make of svc.configd though - over a GB of > phys memory taken by it (and that is on a 2GB machine). I am reluctant to > reboot; this, after all, may repeat itself. > > There are no problem services (svcs -xv returns nothing). > > Does anybody else have similar problem? This is probably http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6765749 although perhaps some change appears to have made this outcome more common. You should be able to pkill svc.configd; svc.startd will start a new one. - Stephen -- s...@sun.com http://blogs.sun.com/sch/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] svc.configd leaking on snv_129?
> * Chavdar Ivanov [2009-12-15 > 15:04]: > > Yesterday I upgraded three of my OpenSolaris > systems to 129; this afternoon I came to find one of > them very sluggish; it turned out that: > > > > PID USERNAME SIZE RSS STATE PRI NICE > TIME CPU PROCESS/NLWP > 11 root 1536M 1048M sleep 590 1:09:27 > 6.3% svc.configd/21 > 4034 postgres 154M 135M sleep 590 > 0:00:01 0.0% postgres/1 > 3917 postgres 154M 135M sleep 590 0:00:00 > 0.0% postgres/1 > 4027 postgres 154M 135M sleep 590 > 0:00:07 0.0% postgres/1 > 4028 postgres 154M 134M sleep 590 0:00:05 > 0.0% postgres/1 > (the rest were less significant). > > I disabled postgres for the moment and got the > system to almost normal responsiveness. Don't know > what to make of svc.configd though - over a GB of > phys memory taken by it (and that is on a 2GB > machine). I am reluctant to reboot; this, after all, > may repeat itself. > > There are no problem services (svcs -xv returns > nothing). > > Does anybody else have similar problem? > This is probably > http://bugs.opensolaris.org/bugdatabase/view_bug.do?b > g_id=6765749 Rather nice analysis here... > > although perhaps some change appears to have made > this outcome more > common. I'll try later to identify the service causing that (pkg/server is not enabled on this box at all). > You should be able to pkill svc.configd; > svc.startd will > start a new one. I thought that, but finding the problem seemed more interesting. > - Stephen > - > s...@sun.com http://blogs.sun.com/sch/ > ___ > opensolaris-discuss mailing list > opensolaris-discuss@opensolaris.org Chavdar Ivanov -- This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] svc.configd leaking on snv_129?
> * Chavdar Ivanov [2009-12-15 > 15:04]: > > Yesterday I upgraded three of my OpenSolaris > systems to 129; this afternoon I came to find one of > them very sluggish; it turned out that: > > > > PID USERNAME SIZE RSS STATE PRI NICE > TIME CPU PROCESS/NLWP > 11 root 1536M 1048M sleep 590 1:09:27 > 6.3% svc.configd/21 > 4034 postgres 154M 135M sleep 590 > 0:00:01 0.0% postgres/1 > 3917 postgres 154M 135M sleep 590 0:00:00 > 0.0% postgres/1 > 4027 postgres 154M 135M sleep 590 > 0:00:07 0.0% postgres/1 > 4028 postgres 154M 134M sleep 590 0:00:05 > 0.0% postgres/1 > (the rest were less significant). > > I disabled postgres for the moment and got the > system to almost normal responsiveness. Don't know > what to make of svc.configd though - over a GB of > phys memory taken by it (and that is on a 2GB > machine). I am reluctant to reboot; this, after all, > may repeat itself. > > There are no problem services (svcs -xv returns > nothing). The problem turned out to be apache22. I was getting: [Wed Dec 16 16:27:44 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes [Wed Dec 16 16:27:46 2009] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] [Wed Dec 16 16:27:47 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes [Wed Dec 16 16:27:49 2009] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] [Wed Dec 16 16:27:50 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes [Wed Dec 16 16:27:52 2009] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] [Wed Dec 16 16:27:53 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes The system rebooted itself about six hours ago; svc.configd had grown to about half a gigabyte when I notice it a few minutes ago. Disabling apache22 and killing svc.configd restored the normal state of things. Now let's see what's the problem with Apache... > > Does anybody else have similar problem? > This is probably > http://bugs.opensolaris.org/bugdatabase/view_bug.do?b > g_id=6765749 s/probably/definitely/ looking at the frequency of Apache restarts, it's just about what the 6765749 analysis says. > > although perhaps some change appears to have made > this outcome more > common. You should be able to pkill svc.configd; > svc.startd will > start a new one. > - Stephen > - > s...@sun.com http://blogs.sun.com/sch/ > ___ > opensolaris-discuss mailing list > opensolaris-discuss@opensolaris.org Chavdar Ivanov -- This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] svc.configd leaking on snv_129?
Looks like you're hitting: http://bugs.opensolaris.org/bugdatabase/view_bug.do?bug_id=6907454 Chavdar Ivanov wrote: * Chavdar Ivanov [2009-12-15 15:04]: Yesterday I upgraded three of my OpenSolaris systems to 129; this afternoon I came to find one of them very sluggish; it turned out that: PID USERNAME SIZE RSS STATE PRI NICE TIME CPU PROCESS/NLWP 11 root 1536M 1048M sleep 590 1:09:27 6.3% svc.configd/21 4034 postgres 154M 135M sleep 590 0:00:01 0.0% postgres/1 3917 postgres 154M 135M sleep 590 0:00:00 0.0% postgres/1 4027 postgres 154M 135M sleep 590 0:00:07 0.0% postgres/1 4028 postgres 154M 134M sleep 590 0:00:05 0.0% postgres/1 (the rest were less significant). I disabled postgres for the moment and got the system to almost normal responsiveness. Don't know what to make of svc.configd though - over a GB of phys memory taken by it (and that is on a 2GB machine). I am reluctant to reboot; this, after all, may repeat itself. There are no problem services (svcs -xv returns nothing). The problem turned out to be apache22. I was getting: [Wed Dec 16 16:27:44 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes [Wed Dec 16 16:27:46 2009] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] [Wed Dec 16 16:27:47 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes [Wed Dec 16 16:27:49 2009] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] [Wed Dec 16 16:27:50 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes [Wed Dec 16 16:27:52 2009] [warn] Init: Session Cache is not configured [hint: SSLSessionCache] [Wed Dec 16 16:27:53 2009] [emerg] (2)No such file or directory: mod_fcgid: Can't create shared memory for size 12993924 bytes The system rebooted itself about six hours ago; svc.configd had grown to about half a gigabyte when I notice it a few minutes ago. Disabling apache22 and killing svc.configd restored the normal state of things. Now let's see what's the problem with Apache... Does anybody else have similar problem? This is probably http://bugs.opensolaris.org/bugdatabase/view_bug.do?b g_id=6765749 s/probably/definitely/ looking at the frequency of Apache restarts, it's just about what the 6765749 analysis says. although perhaps some change appears to have made this outcome more common. You should be able to pkill svc.configd; svc.startd will start a new one. - Stephen - s...@sun.com http://blogs.sun.com/sch/ ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org Chavdar Ivanov ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] svc.configd leaking on snv_129?
Correct. the fix in http://cr.opensolaris.org/~vrthra/fcgid.conf/ sorted the problem. Chavdar Ivanov -- This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org
Re: [osol-discuss] svc.configd leaking on snv_129?
This fix allowed Apache to start ok for me, but now it seems that no PHP content is served. Anyone got any idea why, or how to fix it? -- This message posted from opensolaris.org ___ opensolaris-discuss mailing list opensolaris-discuss@opensolaris.org