I have added a warning on checking the status of self-heal before running the remove-brick operation. Link to the doc: http://ccs-jenkins.gsslab.brq.redhat.com:8080/job/doc-Red_Hat_Gluster_Storage-3.5-Container_Native_Storage_with_OpenShift_Platform-branch-master/lastSuccessfulBuild/artifact/tmp/en-US/html-single/index.html#idm140465404343952 Please note that all the changes/updates to the section are yet to be merged to the master branch. Moving the bug ON_QA.
Currently we have the following doc section for this bug. =========================================================================== Remove device operation triggers a self-heal in the background. The time taken to complete the self-heal operation is proportional to the data on the removed device. Before you perform the remove device operations, you must ensure that the self-heal operations are complete on all the volumes. Run the following command to obtain the volume names: # oc rsh <gluster_pod_name> Run the following command on each volume to check the self-heal status: # gluster volume heal <volname> info =========================================================================== However, please note that there might be more than one volume configured from a device. so we need to mention here explicitly to check the heal info for all the volumes from the device replaced. so, we should document 2 commands here. 1) A way to find the list of volumes carved out of a device 2) how to check heal info on the volume 3) Repeat the heal info for all the volumes found on step 1.
Determining the list of bricks(hence volumes) which belongs to a device is not possible through heketi-cli in this release. We have therefore asked users to check self heal status of all the volumes. Karthick, what method did you use to determine brick list for a device?
(In reply to Raghavendra Talur from comment #7) > Determining the list of bricks(hence volumes) which belongs to a device is > not possible through heketi-cli in this release. We have therefore asked > users to check self heal status of all the volumes. > > Karthick, what method did you use to determine brick list for a device? I had to login to gluster pods and run heal info for each volume to check there are no pending heals. I couldn't find a better way. One way to find the list of bricks used in a device is to use device info, but it doesn't give the volumes being used. We might have to use topology info further to manipulate volume information. heketi-cli device info 724d4c878d4f406cfeb4bca3bcc15bb0 Device Id: 724d4c878d4f406cfeb4bca3bcc15bb0 Name: /dev/sdd State: online Size (GiB): 99 Used (GiB): 10 Free (GiB): 89 Bricks: Id:1f749f4f5dfab2ee2ac34b9fccbfdbe7 Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_1f749f4f5dfab2ee2ac34b9fccbfdbe7/brick Id:3c78af6a861180417f8763a1fbbaf8e6 Size (GiB):2 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_3c78af6a861180417f8763a1fbbaf8e6/brick Id:5a137d789c0837f93701228d4aea1ce0 Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_5a137d789c0837f93701228d4aea1ce0/brick Id:61493cbf7f58e3233fd1df0a4b2f5093 Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_61493cbf7f58e3233fd1df0a4b2f5093/brick Id:70aa47765d71c4b62328870e704e1506 Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_70aa47765d71c4b62328870e704e1506/brick Id:85a0e27494c162226321e3d48139fcda Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_85a0e27494c162226321e3d48139fcda/brick Id:c1f712413bd9891219deeef98fe3cda9 Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_c1f712413bd9891219deeef98fe3cda9/brick Id:d450ab810010e4a509491beefb4d22e6 Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_d450ab810010e4a509491beefb4d22e6/brick Id:d7624a8b4ffd1312b38dbe855823f221 Size (GiB):1 Path: /var/lib/heketi/mounts/vg_724d4c878d4f406cfeb4bca3bcc15bb0/brick_d7624a8b4ffd1312b38dbe855823f221/brick
As Karthick has mentioned in previous comment, it is not possible to determine volume list corresponding to a device. Volume info command does not have sufficient information either. Hence the old doc which says that user must check for self heal status of all the volumes remains the right way.
Based on Comment 9, moving back the bug ON_QA.
Looks good to me, moving the bug to verified.