Hide Forgot
Document URL: https://access.redhat.com/documentation/en-US/Red_Hat_Storage/3.1/html/Administration_Guide/chap-Managing_Data_Tiering-Limitations.html Section Number and Name: 12.3. Tiering Limitations Describe the issue: 12.3. Tiering Limitations ... When you run tier detach commit force, ongoing I/O operation may fail with Transport endpoint is not connected error. ... ----- Please note that the command can be either of: 1. tier detach commit OR 2. tier detach force tier detach commit force - is not a valid gluster cli command ----- You might want to represent it as: tier detach { commit | force }
Thanks for creating the BZ. There's still some engineering debate going on what the command should actually be. So, I'll comment on this patch accordingly.
The following things have been confirmed: # gluster volume tier <VOLNAME> attach [force] The tier attach command can accept an optional force argument. The force argument is required when re-attaching bricks which have been detached earlier. During the detach, some configuration bits are left behind that are detected during the re-attach and the attach is aborted. Specifying the force argument assumes that the warning presented during the earlier attach attempt without the force argument has been read and understood by the user and the user knows the significance of using the force argument. # gluster volume tier <VOLNAME> detach commit The tier detach commit command does NOT accept any force argument. The tier detach commit command is supposed to be used after the tier detach start command to commit the detach operation. # gluster volume tier <VOLNAME> detach force The tier detach command accepts a force command on its own to force a detach.
Laura, I'm presuming this is a question regarding 'tier <vol> attach' command. When bricks are detached and attempted for a re-attach, you'll get to see a message stating that the 'brick is already part of a volume'. When a brick is detached, the extended attributes are not removed from the detached brick(s). On a re-attach attempt, this 'stale' information is detected and a message I mentioned above is spit out. The reason that the message is spit out is to raise a flag to the user as a word of caution. If the user is confident her actions, then the 'force' argument can be passed to the 'tier <vol> attach' command to re-attach the brick(s) to the volume. Please note that argument that needs to be passed is 'force' and not '--force'
perfect
Have shared my review comments with Laura over an email.
verified the bz on document mentioned in the comment 14. Hence moving to verified
Moving to CLOSED CURRENTRELEASE since RHGS 3.2 GA was yesterday. All documentation is available from https://access.redhat.com/documentation/en/red-hat-gluster-storage/.