[blivet] refactor lvm info gathering

David Lehman dlehman at redhat.com
Fri Mar 28 20:32:49 UTC 2014


This slightly improves how we gather information from lvm about
existing devices. It then adds caching to consolidate calls to lvm
utilities on systems with many PVs and/or VGs.

This version does not enable caching for anaconda, but I think I
should remove that piece. Does anyone object? What it means is that
we'll only run pvs and lvs once per call to DeviceTree.populate,
regardless of the number of PVs or VGs present. I think we call
populate after any operation that could add a new PV in anaconda, so
it should be safe.

Jan, if you're reading this, I'd also be open to adding a way to tell
the devicetree not to drop the cache on populate. Let me know if that
sounds useful.


More information about the anaconda-patches mailing list