Jonathan C. Detert wrote:
If there's corruption that was not between the tail to the head,
then it could be a goer I guess but one would wonder about corruption
what do you mean by 'goer' - hardware failure?
I was just thinking that if there was corruption just confined to the
log but not in the interval between the tail and head then
I can't see why theoretically the log can't be replayed.
The replay of the log is done from the tail to the head - the
other log data is mostly irrelevant in this case AFAICT
(I think the failure came with code which is trying to deal
with the case with a new filesystem where part of the log
is not written to yet so has zero cycle#s - but here we have
wrapped 598 times or whatever it was).
--Tim
|