Dan Kenigsberg has posted comments on this change.
Change subject: Invalidate filters on HSMs before rescanning extended VG ......................................................................
Patch Set 18:
It's hard to have a multi-participant conversation in gerrit, but I'll try to join.
Sergey and Ayal: The devices may be visible in getDevVis, but clearly there has been a time in the past where they did not exist (that's the reason of this bug, devices popping up unattended).
If the domainMonitor has issued its vgs command when the new lun was not there, it's filter would have lacked the new device. On other threads, getDevVis may now report the device as visible, a vgextend command takes place, and the old domainMonitor would return a dreaded "partial" result.
Federico: I believe that connectStorageServer is not sent for FC, which is the core of the bug. If it was called after the lun was visible on HSM, its invalidateStorage() call would have given HSM to rebuild its filter.
Nir: as Ayal said, this 3.2 bug is about the "long" filter, not the new per-vg short filter.