[OpenAFS-devel] volume header timestamps

[Benjamin Kaduk]

On Mon, 26 Jan 2015, Benjamin Kaduk wrote:

> and summarize some of the recent topics of discussion in separate
> messages, to make the threading easier if discussion crops up on -devel.

There have long been some ~edge cases with how the various timestamps in
the volume header are handled on dump, clone, restore, etc..

The change in gerrit 11389 was merged back in August, but exposed another
issue, believed to be resolved by 11468, with 11627 on top for good
measure.  The latter two changes are sitting in gerrit with CodeReview=+2
but have not yet been merged.

I believe jhutz has sent some analysis of the relevant issues to this list
previously, so I won't try to recap right now.  In any case, there seems
to be agreement that 11468+11627 is a definite improvement, so we will
probably go with them, and defer discussion about whether volume metadata
changes should update the header timestamps, as opposed to only volume
data updates as is the current state.

-Ben