Ayal Baron has posted comments on this change.
Change subject: Invalidate filters on HSMs before rescanning extended VG ......................................................................
Patch Set 18:
Dan, clearly there is not enough clarity on the flow which is leading you to wrongly think there is a race here. The flow is: on SPM host: discover and connect to storage (optional) getDeviceList in GUI choose a new device to add to domain
Now *before* vgextend is run, on *all* hosts: discover and connect to storage (if needed) getDevicesVisibility
Only after all hosts have reported that the new device(s) is visible (and filter is up to date) do we go back to the spm and run: vgextend
On top of all this, a host only moves to non-op after *5* minutes !!! (to differentiate between host and total storage failure).
So now is it clear that there is no race?