[ovs-dev] 100% CPU in ovs-vswitchd (again)

Ben Pfaff blp at nicira.com
Tue Sep 30 16:37:16 UTC 2014


There's still a 100% CPU bug in ovs-vswitchd.  Run "make sandbox", then
run a second ovs-vswitchd inside the subshell (just run a bare
"ovs-vswitchd" without any parameters).  You get output like this:

    2014-09-30T16:33:45Z|00001|ovs_numa|INFO|Discovered 8 CPU cores on NUMA node 0
    2014-09-30T16:33:45Z|00002|ovs_numa|INFO|Discovered 1 NUMA nodes and 8 CPU cores
    2014-09-30T16:33:45Z|00003|reconnect|INFO|unix:/home/blp/nicira/ovs/tutorial/sandbox/db.sock: connecting...
    2014-09-30T16:33:45Z|00004|reconnect|INFO|unix:/home/blp/nicira/ovs/tutorial/sandbox/db.sock: connected
    2014-09-30T16:33:45Z|00005|bridge|ERR|another ovs-vswitchd process is running, disabling this process (pid 13812) until it goes away
    2014-09-30T16:33:48Z|00006|poll_loop|INFO|wakeup due to 1-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00007|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00008|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00009|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00010|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00011|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00012|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00013|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00014|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)
    2014-09-30T16:33:48Z|00015|poll_loop|INFO|wakeup due to 0-ms timeout at ../vswitchd/bridge.c:2723 (100% CPU usage)

The cited line in bridge.c is:

        seq_wait(connectivity_seq_get(), connectivity_seqno);

Thanks,

Ben.



More information about the dev mailing list