Description of problem: Echo request from OVS uses default value today, which is 5 seconds. In scale setups, Full GC can take more than 5 seconds which will result in a disconnection from all OVSs. To control this behavior, inactivity_probe can be increased when creating a bridge. OVSDB already created a patch that will allow its clients to define their desired inactivity_probe: https://git.opendaylight.org/gerrit/#/c/50643/ Now Netvirt needs to expose an ability to configure this externally.
Proposed the following downstream patch: https://code.engineering.redhat.com/gerrit/#/c/98244/
Hi, Please specify how to verify.
Steps to verify: Create the file “etc/opendaylight/datastore/initial/config/netvirt-elanmanager-config.xml” with the following params before starting karaf. <elanmanager-config xmlns="urn:opendaylight:netvirt:elan:config"> <auto-create-bridge>true</auto-create-bridge> <int-bridge-gen-mac>true</int-bridge-gen-mac> <temp-smac-learn-timeout>10</temp-smac-learn-timeout> <punt-lldp-to-controller>false</punt-lldp-to-controller> <controller-max-backoff>10000</controller-max-backoff> <controller-inactivity-probe>300000</controller-inactivity-probe> </elanmanager-config> Now, start the controller and continue with stacking. On one of the compute nodes look at the "inactivity_probe" and "max_backoff" configured in the OVSDB Controller table. [vagrant@f23_aione_newton_odl_setup3 ~]$ sudo ovsdb-client dump unix:/var/run/openvswitch/db.sock Controller inactivity_probe max_backoff Controller table inactivity_probe max_backoff ---------------- ----------- 300000 10000
Checked with version: opendaylight-5.2.0-6.el7ost.noarch I changed the file /opt/opendaylight/etc/opendaylight/datastore/initial/config/netvirt-elanmanager-config.xml during installation with the following content: <elanmanager-config xmlns="urn:opendaylight:netvirt:elan:config"> <controller-max-backoff>10000</controller-max-backoff> <controller-inactivity-probe>400000</controller-inactivity-probe> </elanmanager-config> Then ran on a compute node: $ sudo ovsdb-client dump unix:/var/run/openvswitch/db.sock Controller inactivity_probe max_backoff And got: Controller table inactivity_probe max_backoff ---------------- ----------- 300000 10000
Since the problem described in this bug report should be resolved in a recent advisory, it has been closed with a resolution of ERRATA. For information on the advisory, and where to find the updated files, follow the link below. If the solution does not work for you, open a new bug report. https://rhn.redhat.com/errata/RHBA-2017-0840.html