swift-drive-audit should handle drives that have physically or logically disappeared

Bug #1266552 reported by david
This bug affects 1 person
Affects Status Importance Assigned to Milestone
OpenStack Object Storage (swift)

Bug Description

On a object server, if an operator removes the wrong drive by accident, the disk remains mounted from the OS perspective.

The swift-drive-audit could handle this sort of problem because the disk is removed from /proc/partitions but currenlly it does not unmount the disk

The regex will likely match the error messages the OS or disk controller produces, but as an entry in /proc/partitions is now gone, the run of swift-drive-audit will not be able to match the major/minor pair to a /proc/partitions line and it therefore won't try to unmount the disk.

This can also happen if a disk logically fails - we see occasion disk locks ups whick do not get caught by swift-drive-audit also.

Revision history for this message
Filippo Giunchedi (filippo) wrote :

https://bugs.launchpad.net/swift/+bug/1359664 partly handles this problem btw

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.