[Info-vax] Re; Spiralog, RMS Journaling (was Re: FREESPADRIFT)

[Michael Moroney]

lawrencedo99 at gmail.com writes:

>On Saturday, June 18, 2016 at 2:26:45 AM UTC+12, Stephen Hoffman wrote:

>Then where is the inconsistency in free-space usage coming from? You 
>know, the issue that started this thread?

I can't tell for certain without more info, but in a cluster wide mount,
some stuff like file headers and disk space bitmaps are allocated to each
cluster member (and marked as allocated on the disk) so the cluster member
can create/extend files without all the inter-cluster traffic required to
update cluster members every time this happens.  When a disk is properly
dismounted, unused allocated space is returned.  If a cluster member
crashes, it is possible for other members to not be able to mark space
allocated to the departed member free, it remains as a ghost allocation.  
MOUNT/REBUILD normally clears this stuff up.