[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [Xen-users] Weird ram/dom-u limit problem
Hello all,I've been having a very odd problem the last few days and I'm at a loss on how to proceed, so I thought I'd ask the user community. I'm hitting what *appears* to be some kind of wall on either the number of vm's or the amount of ram they're using (personally I think it's ram, but I don't see how). We have two identical CentOS 5.4 xen servers (xen-3.0.3-94.el5_4.2) in production with 7 vm's and dom0 on each server. I'll post the full config on the bottom of the email, but here are the basics. xentop - 13:58:32 Xen 3.1.2-164.6.1.el5 8 domains: 1 running, 7 blocked, 0 paused, 0 crashed, 0 dying, 0 shutdown Mem: 16776420k total, 5582324k used, 11194096k free CPUs: 4 @ 3000MHz dom0_mem=1g loops = 64Now, if I attempt to install an 8th vm with 512MB or ram on either server the install runs for awhile and then suddenly load on the xen server begins to grow and grow and grow. Going from a load average: 0.09, 0.15, 0.10 to load average: 10.58, 7.88, 4.70. The box itself begin to become very slow and bogged down. top shows that it's because of cpu iowait, but I can't find any particular reason for it. top - 13:35:18 up 63 days, 13:54, 3 users, load average: 10.58, 7.88, 4.70 Tasks: 196 total, 1 running, 195 sleeping, 0 stopped, 0 zombieCpu(s): 0.0%us, 0.0%sy, 0.0%ni, 24.9%id, 74.5%wa, 0.0%hi, 0.0%si, 0.7%st 75% is ridiculously high and it continues to get worse until the install finally barely finishes after much longer than normal. An average install takes 20 minutes, they're taking around 35-40 now and I fear that if I add yet another vm it will be even worse. Once the install finishes the box returns to normal and I can boot the new vm normally as well. If I shutdown one of the existing vm's using 512MB of ram, then the install works perfectly and the load never goes crazy. Turn the 7th back on , and installing the 8th is horrible again. I can duplicate this on both servers which is very weird, but I can't figure out *why* it's happening. The fact that both servers have the problem leads me to believe its a config problem. I'm leaning towards something running out of ram and swapping hard, which is why load takes awhile to start climbing and the cpu wait gets very high, yet I have *more* than enough memory allocated to everything. The installs are very io intensive, but I can't understand why that 8th vm is causing hell. Is there some weird limit to the amount of vm's or ram I can allocate? Does dom0 need more than 1g and that's why it's getting cranky? Is there a config option I'm not aware of? Any help would be appreciated, this is really worrying me. Thanks. Swap: # free total used free shared buffers cached Mem: 1048576 1039176 9400 0 10924 777816 -/+ buffers/cache: 250436 798140 Swap: 10482404 188 10482216 Full config: # xm info host : xen2.rutgers.edu release : 2.6.18-164.6.1.el5xen version : #1 SMP Tue Nov 3 16:48:13 EST 2009 machine : x86_64 nr_cpus : 4 nr_nodes : 1 sockets_per_node : 2 cores_per_socket : 2 threads_per_core : 1 cpu_mhz : 3000hw_caps : 178bfbff:ebd3fbff:00000000:00000010:00002001:00000000:0000001f total_memory : 16383 free_memory : 10931 node_to_cpu : node0:0-3 xen_major : 3 xen_minor : 1 xen_extra : .2-164.6.1.el5xen_caps : xen-3.0-x86_64 xen-3.0-x86_32p hvm-3.0-x86_32 hvm-3.0-x86_32p hvm-3.0-x86_64 xen_pagesize : 4096 platform_params : virt_start=0xffff800000000000 xen_changeset : unavailable cc_compiler : gcc version 4.1.2 20080704 (Red Hat 4.1.2-46) cc_compile_by : mockbuild cc_compile_domain : centos.org cc_compile_date : Tue Nov 3 16:04:14 EST 2009 xend_config_format : 2 -- ================================ David Halik System Administrator OIT-CSS Rutgers University dhalik@xxxxxxxxxxxxxxx ================================ _______________________________________________ Xen-users mailing list Xen-users@xxxxxxxxxxxxxxxxxxx http://lists.xensource.com/xen-users
|
Lists.xenproject.org is hosted with RackSpace, monitoring our |