Re: [Users] management server very slow lately

On 03/22/2013 02:54 PM, Jonathan Horne wrote: > top - 08:53:38 up 70 days, 16:31, 1 user, load average: 0.40, 0.34, >0.32 > Tasks: 432 total, 1 running, 431 sleeping, 0 stopped, 0 zombie > Cpu(s): 1.3%us, 0.1%sy, 0.0%ni, 98.6%id, 0.0%wa, 0.0%hi, 0.0%si, >0.0%st > Mem: 32876240k total, 18653508k used, 14222732k free, 522432k buffers > Swap: 2097144k total, 4528k used, 2092616k free, 6270908k cached > > PID USER PR NI VIRT RES SHR S %CPU %MEM TIME+ COMMAND > 2121 ovirt 20 0 12.9g 7.7g 18m S 9.0 24.6 16539:08 java > This is not normal at all. First thing that is strange is that your engine is taking 7.7 GiB of RAM, which it should never take, as it is by default limited to 1 GiB. Did you assign more memory to the engine on purpose? How much? If you assign a lot of memory it can start to consume a lot of CPU just for garbage collection. You may want to enable verbose garbage collection adding this to /etc/sysconfig/ovirt-engine (or /etc/ovirt-engine/engine.conf if you are using the latest source code): ENGINE_VERBOSE_GC=true Then restart the engine and it will start to dump garbage collection statistics to /var/log/ovirt-engine/console.log. The garbage collection should be quite silent in an low activity system. We used to have a bug that caused the max amount of memory not be correctly limited, but it was fixed long ago: http://gerrit.ovirt.org/7952 The other thing that seems strange is the amount of CPU that it is consuming. Do you have many hosts managed by that engine? In an otherwise idle environment the CPU consumption is caused by the periodic polls of the hosts, one each two seconds by default. If you see continually the engine using a significant amount of CPU (you the output of top above it is 9%) it could be useful to get a snapshot of the stacks of threads, to see which threads in particular are consuming the CPU. Send the QUIT signal to the engine process and it will dump the stacks of the threads to /var/log/ovirt-engine/console.log: # kill -3 $(cat /var/run/ovirt-engine.pid) Once you have that dump you can check which thread is consuming the CPU as follows: 1. Get the PIDs of the threads of the engine together with their use of CPU: # ps -L -u ovirt -o tid,pcpu 2. If you see one of them consuming a high amount of CPU time then try to find it in the stack dump generated in /var/log/ovirt-engine/console.log. Lets assume that the PID is 13397, for example, translate it to hex: # printf "%04x\n" 13397 3455 3. Then look in /var/log/ovirt-engine/console.log for a line containing "nid=0x3455". There you will find the stack trace of that thread, something like this: "ajp-/127.0.0.1:8702-Acceptor-0" daemon prio=10 tid=0x00007f41e0220800 nid=0x3493 runnable [0x00007f41dbdf2000] java.lang.Thread.State: RUNNABLE ... Most threads will be waiting, but if you find one thread that is consistently RUNNABLE then there is probably an issue. The dump of the stack of that thread can help to find out what it is doing and why it is consuming the CPU. > > I don't have a lot of experience with jboss, so im not sure it thats >good or bad. I did the jboss restart, and that helped a little, but its >still a little sluggish again, now a few days later. > > Thanks, > > -----Original Message----- > From: Itamar Heim [mailto:iheim@redhat.com] > Sent: Friday, March 15, 2013 6:32 AM > To: Jonathan Horne > Cc: users(a)ovirt.org > Subject: Re: [Users] management server very slow lately > > On 03/13/2013 08:51 PM, Jonathan Horne wrote: >> Hello, lately my manager server web interface is extremely sluggish. >> Perhaps the server is ready for a reboot? >> >> My management server is also the hosts of my NFS export and ISO mounts. >> Is there a prescribed method for rebooting when I am also providing >> NFS services from the management server? My assumption is that aside >> from NFS, I should be able to reboot the management serve and the >> nodes and virtual machines will be fine in the mean time? > > what's the cpu consumption of your ovirt-engine service (java process). > cpu load on the engine? memory/swap state of the engine, etc > > > ________________________________ > This is a PRIVATE message. If you are not the intended recipient, >please delete without copying and kindly advise us by e-mail of the >mistake in delivery. NOTE: Regardless of content, this e-mail shall not >operate to bind SKOPOS to any order or other contract unless pursuant to >explicit written agreement or government initiative expressly permitting >the use of e-mail for such purpose. > _______________________________________________ > Users mailing list > Users(a)ovirt.org > http://lists.ovirt.org/mailman/listinfo/users > -- Dirección Comercial: C/Jose Bardasano Baos, 9, Edif. Gorbea 3, planta 3ºD, 28016 Madrid, Spain Inscrita en el Reg. Mercantil de Madrid ­ C.I.F. B82657941 - Red Hat S.L.