Document URL: ============= https://access.qa.redhat.com/documentation/en-us/red_hat_openshift_container_storage/3.10/html-single/deployment_guide/#chap-Documentation-Red_Hat_Gluster_Storage_Container_Native_with_OpenShift_Platform-Upgrade-Gluster_pods Section Number and Name: ============================ Chapter 6. Upgrading your Red Hat Openshift Container Storage in Converged Mode Describe the issue: ===================== Customers/Teams are facing some issues in interpreting the exact changes for upgrades in the heketi, glusterfs and gluster-block-prov templates. See - https://bugzilla.redhat.com/show_bug.cgi?id=1629728#c13 The deployment guide has steps to upgrade glusterfs pods and heekti pods . But the steps are generalized for both glusterfs and glusterfs-registry namespaces. It would help if a separate section for the changes required in glusterfs-registry are also elaborately explained. E.g. 6.3. Upgrading cns-deploy and Heketi Server Step 4 says - "Update the heketi template to change the HEKETI_ROUTE, IMAGE_NAME, IMAGE_VERSION and CLUSTER_NAME" But we would need proper snips of all the 4 variables for both glusterfs and glusterfs-registry as the values will be different. In a nutshell, we may need the changes specified by Daniel - https://bugzilla.redhat.com/show_bug.cgi?id=1629728#c13 Suggestions for improvement: Additional information:
Hello Bhavana, I have some review comments below. 1) I do not see the steps to bring down heketi pod before upgrading glusterfs server pods 2) In point 10 of section upgrading glusterfs pods, command for checking heal info on the volume needs to be changed as the provided command as of today only checks for one volume. 3) Note from step 12 has to be removed which talks about upgrading glusterfs registry pods as we have a separate section to do that. 4) After step 13 another step needs to be added which talks about rebooting gluster pods after upgrading of op-version. 5) Example in point no.15 has to be changed as # oc delete dc glusterblock-storage-provisioner-dc 6) Comments provided here for glusterfs_registry https://bugzilla.redhat.com/show_bug.cgi?id=1644200#c11 Moving this bug back to assigned state to address the above comments.
@Neha, At point no.5 should not we say oc get all | grep heketi to get all names of these parameters ?
(In reply to RamaKasturi from comment #12) > @Neha, > > At point no.5 should not we say oc get all | grep heketi to get all names of > these parameters ? Hi kasturi, Its ok to have a 'oc get all|grep heketi' instead of the generic "oc get all". During dicsussion, we may have thought to keep the generic command and then based on the outputs, delete the respective heketi, gluster & glusterblock-provisioner parameters in subsequent steps. Thanks, Neha
From the bugzilla it looks like these changes have to be made to 3.10 docs where as the link provided here is 3.11. So moving the bug back to assigned state.
Acking the bug for 3.11.3 since i see that there is no section yet for upgrading glusterfs_registry pods in 3.10 guides.