Description of problem: The OCS upgrade playbook misses updating the multipath.conf file. As part of OCS upgrades, we must update the multipath.conf file after all the server nodes are upgraded. Once updating the multipath.conf file is successful, we should start and reload multipathd. file to be updated: /etc/multipath.conf ------- below snippet as per the 3.11 doc ---------- # cat >> /etc/multipath.conf <<EOF # LIO iSCSI devices { device { vendor "LIO-ORG" user_friendly_names "yes" # names like mpatha path_grouping_policy "failover" # one path per group hardware_handler "1 alua" path_selector "round-robin 0" failback immediate path_checker "tur" prio "alua" no_path_retry 120 } } EOF ---------- /End of Snippet ---------------------- Version-Release number of selected component (if applicable): All OCS versions Actual results: The OCS upgrade playbook misses updating multipath.conf file after server nodes are upgraded. Expected results: The OCS upgrade playbook should update the multipath.conf file after server nodes are upgraded and then start and reload multipathd
PR is upstream: https://github.com/openshift/openshift-ansible/pull/11515
PR is merged.
Hello Jose, Can you please update the FIV for this bug ? Thanks kasturi
FIV - Fixed In version. Can you update the ansible package which has this bug fixed, in Fixed In Version of the Bugzilla field.
New PR is upstream: https://github.com/openshift/openshift-ansible/pull/11609
PR is merged. Fix will be available in the next build which should be tonight.
To verify: As part of upgrade, a new section for device with vendor value as "LIO-ORG" will be added. You can check /etc/multipath.conf before upgrade and ensure "LIO-ORG" added after upgrade. Refer: https://github.com/openshift/openshift-ansible/blob/release-3.11/roles/openshift_node/tasks/storage_plugin_iscsi.yml#L59 device { vendor "LIO-ORG" product "TCMU device" user_friendly_names "yes" path_grouping_policy "failover" path_selector "round-robin 0" failback immediate path_checker "tur" prio "alua" hardware_handler "1 alua" no_path_retry 120 rr_weight "uniform" }
Confirming the same as Ashmitha wrote in comment 21, That we should verify the values of prio etc.
(In reply to Nitin Goyal from comment #22) > Confirming the same as Ashmitha wrote in comment 21, That we should verify > the values of prio etc. I think there is some miscommunication here. Updating new section for having gluster-block configured is the main bz here.(Ref comment #0) There are many different values here as part of device (like prio, rr_weight, no_path_retry etc.,). It is not possible to check for all the different values present as part of device section. The original PR posted by Jose is to include this task(iscsi.yml) as part of upgrade playbook: Refer: https://github.com/openshift/openshift-ansible/pull/11515/files#diff-d4673ae0996f46748c28b94ecdb4169dR13 from https://github.com/openshift/openshift-ansible/pull/11515