On Thu, Oct 10, 2013 at 8:49 AM, Sahid Ferdjaoui <sahid.ferdja...@cloudwatt.com> wrote: > Ok thanks for your clarification, i'm going to work on that... > Just, do you know if there is a package available for ubuntu 12.04 (even > tests purposes) or do i have to build one?
I am not aware of a place where you can download packages like that. The INSTALL.Debian file explains how to create packages for Ubuntu. > > ----- Original Message ----- > From: "Gurucharan Shetty" <shet...@nicira.com> > To: "Sahid Ferdjaoui" <sahid.ferdja...@cloudwatt.com> > Cc: "discuss" <discuss@openvswitch.org> > Sent: Thursday, October 10, 2013 5:21:25 PM > Subject: Re: [ovs-discuss] Tune for large number of connections... > > On Thu, Oct 10, 2013 at 8:07 AM, Sahid Ferdjaoui > <sahid.ferdja...@cloudwatt.com> wrote: >> Ok, does it means the server can handle more connections? > > In 1.11, there is kernel wildcarding for flows. So if you are using > OVS as a simple mac learning switch, then flows with different TCP > ports are not treated as separate flows but rather a single flow. So, > a single flow in kernel will be good enough to handle all your TCP > connections. My answer is based on the assumption that you do not have > openflow flows that want to look at every TCP port and do something > about it. > >> >> ----- Original Message ----- >> From: "Gurucharan Shetty" <shet...@nicira.com> >> To: "Sahid Ferdjaoui" <sahid.ferdja...@cloudwatt.com> >> Cc: "discuss" <discuss@openvswitch.org> >> Sent: Thursday, October 10, 2013 4:46:34 PM >> Subject: Re: [ovs-discuss] Tune for large number of connections... >> >> On Thu, Oct 10, 2013 at 2:00 AM, Sahid Ferdjaoui >> <sahid.ferdja...@cloudwatt.com> wrote: >>> Hello all, >>> >>> I have a vm1 (server) on a hostA uses ovs and some other vmX (client) on >>> hostB or hostC... >>> >>> When the number of connections go up than about 53000 connections (socket >>> tcp) I get a lot of errors: >>> The deal is that, there are enough memory and cpu and if I try a test >>> without ovs (I mean with a physical network) I can easily go up this limit. >>> >>> version: 1.10.1 >> 1.11 shouldn't show these issues. >>> >>> do you have an idea for me? >>> Thanks a lot >>> s. >>> >>> 2013-10-10T08:55:10Z|00475|dpif_linux|WARN|system@ovs-system: lost packet >>> on channel 9 (last polled 0 ms ago) >>> 2013-10-10T08:55:11Z|00476|dpif_linux|WARN|system@ovs-system: lost packet >>> on channel 9 (last polled 0 ms ago) >>> 2013-10-10T08:55:11Z|00477|ofproto_dpif_governor|INFO|br-int: processed >>> 32768 packets in 0.43 s, enlarging hash table to 256 kB (46522 hashes, 260 >>> setups, 0 shortcuts) >>> 2013-10-10T08:55:20Z|00478|ofproto_dpif_governor|INFO|br-int: processed >>> 34824 packets in 5.00 s, shrinking hash table to 128 kB (45536 hashes, 1059 >>> setups, 0 shortcuts) >>> 2013-10-10T08:55:20Z|00479|dpif_linux|WARN|system@ovs-system: lost packet >>> on channel 9 (last polled 0 ms ago) >>> 2013-10-10T08:55:21Z|00480|ofproto_dpif_governor|INFO|br-int: processed >>> 32768 packets in 0.53 s, enlarging hash table to 256 kB (43473 hashes, 537 >>> setups, 0 shortcuts) >>> 2013-10-10T08:55:21Z|00481|dpif_linux|WARN|system@ovs-system: lost packet >>> on channel 9 (last polled 0 ms ago) >>> 2013-10-10T08:55:21Z|00482|dpif_linux|WARN|system@ovs-system: lost packet >>> on channel 9 (last polled 0 ms ago) >>> 2013-10-10T08:55:22Z|00483|poll_loop|WARN|Dropped 26318 log messages in >>> last 59907 seconds (most recently, 59904 seconds ago) due to excessive rate >>> 2013-10-10T08:55:22Z|00484|poll_loop|WARN|Dropped 3 log messages in last >>> 59919 seconds (most recently, 59907 seconds ago) due to excessive rate >>> 2013-10-10T08:55:22Z|00485|poll_loop|WARN|wakeup due to 0-ms timeout at >>> ../ofproto/ofproto-dpif.c:1174 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00486|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00487|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00488|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00489|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00490|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00491|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00492|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00493|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:22Z|00494|poll_loop|WARN|wakeup due to [POLLIN] on fd 19 >>> (unknown anon_inode:[eventpoll]) at ../lib/dpif-linux.c:1363 (51% CPU usage) >>> 2013-10-10T08:55:28Z|00495|ofproto_dpif_governor|INFO|br-int: processed >>> 57094 packets in 5.00 s, shrinking hash table to 128 kB (49357 hashes, 3265 >>> setups, 0 shortcuts) >>> 2013-10-10T08:55:31Z|00496|dpif_linux|WARN|system@ovs-system: lost packet >>> on channel 9 (last polled 0 ms ago) >>> 2013-10-10T08:55:31Z|00497|ofproto_dpif_governor|INFO|br-int: processed >>> 32768 packets in 0.43 s, enlarging hash table to 256 kB (45883 hashes, 837 >>> setups, 0 shortcuts) >>> ... >>> >>> >>> >>> >>> _______________________________________________ >>> discuss mailing list >>> discuss@openvswitch.org >>> http://openvswitch.org/mailman/listinfo/discuss _______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss
