On Wed, Feb 20, 2019 at 10:13 PM Lars Kellogg-Stedman <lars@redhat.com> wrote:
I was trying to track down some connectivity issues with some baremetal nodes booting from iSCSI LUNs provided by Cinder. It turns out that openvswitch is going belly-up. We see in openvswitch-agent.log a "Switch connection timeout" error [1]. Just before that, in /var/log/openvswitch/ovs-vswitchd.log, we see:
2019-02-21T00:32:38.696Z|00795|bridge|INFO|bridge br-tun: deleted interface patch-int on port 1 2019-02-21T00:32:38.696Z|00796|bridge|INFO|bridge br-tun: deleted interface br-tun on port 65534 2019-02-21T00:32:38.823Z|00797|bridge|INFO|bridge br-int: deleted interface int-br-ctlplane on port 1 2019-02-21T00:32:38.823Z|00798|bridge|INFO|bridge br-int: deleted interface br-int on port 65534 2019-02-21T00:32:38.823Z|00799|bridge|INFO|bridge br-int: deleted interface tapb0101920-b9 on port 4 2019-02-21T00:32:38.824Z|00800|bridge|INFO|bridge br-int: deleted interface patch-tun on port 3 2019-02-21T00:32:38.954Z|00801|bridge|INFO|bridge br-ctlplane: deleted interface phy-br-ctlplane on port 4 2019-02-21T00:32:38.954Z|00802|bridge|INFO|bridge br-ctlplane: deleted interface br-ctlplane on port 65534 2019-02-21T00:32:38.954Z|00803|bridge|INFO|bridge br-ctlplane: deleted interface em2 on port 3
The plot thickens: it looks as if something may be doing this explicitly? At the same time, we see in the system journal: Thu 2019-02-21 00:32:38.697531 UTC [s=fa1c368ed0314169b286a29ffe7d9f87;i=4aa2b;b=d64947ee218546d8a94103aa9bbee154;m=4b31e6780;t=5825c9c8dda3b;x=13507b4516ddcb27] _TRANSPORT=stdout PRIORITY=6 SYSLOG_FACILITY=3 _UID=0 _GID=0 _CAP_EFFECTIVE=1fffffffff _SELINUX_CONTEXT=system_u:system_r:init_t:s0 _BOOT_ID=d64947ee218546d8a94103aa9bbee154 _MACHINE_ID=4a470fefdd3b4033a163bb69bc8578da _HOSTNAME=localhost.localdomain _SYSTEMD_SLICE=system.slice _EXE=/usr/bin/bash _STREAM_ID=951219afe7fb4eb4bf5d71af983d1f11 SYSLOG_IDENTIFIER=ovs-ctl MESSAGE=Exiting ovs-vswitchd (9340) [ OK ] _PID=868225 _COMM=ovs-ctl _CMDLINE=/bin/sh /usr/share/openvswitch/scripts/ovs-ctl --no-ovsdb-server stop _SYSTEMD_CGROUP=/system.slice/ovs-vswitchd.service/control _SYSTEMD_UNIT=ovs-vswitchd.service ...but there's nothing else around that time that seems relevant. -- Lars Kellogg-Stedman <lars@redhat.com>