Nir Soffer has posted comments on this change.
Change subject: Invalidate filters on HSMs before rescanning extended VG ......................................................................
Patch Set 18:
What happens is this:
1. In getDevicesVisibility, I can see the new device (but it is not in the filter, because we do not update it normally) 2. We now mark the filter as stale (this patch) 3. SPM extends the vg with new device
What we like to happen, is that the filter is updated after 3 by the next lvm command (filters are built when a command is run and filter is stale)
What may happen (the race Eduardo is talking about), is that filter is updated before 3, when the vg is not yet used by the vg.
So yes there is little race, but it does not cause any harm, since the filter always contains all visible devices on the machine.
So this patch is good enough for 3.2 and 3.3 and should be merged *now*.
Should be have even better way to do it in 3.4? maybe - not interesting now.