Simon Wilkinson <[email protected]> writes: > On 9 Jan 2010, at 21:34, Russ Allbery wrote:
>> Yes, although something that would make this sort of thing easier in the >> long run may also be of interest (it just can't be the entire solution). > Would making fs flushvolume not return control to the user until the > data has been removed from the cache be sufficient? This would actually > be fairly straightforward to implement. Well, we don't necessarily know what volume the data was in. It would be nice to have fs flushall for UNIX as well. But the combination of those two things would solve the problem, I think. > A further consideration is the level of destruction of the cached copies > that you require. At the moment, all the cache manager will ever do is > truncate the cache file to 0 bytes. Yeah, that's a good point, but for our immediate purposes we're going after equivalent destruction as deleting a file. I realize that may not make the data go away entirely; it's mostly a matter of being at least as good as what we have now. -- Russ Allbery ([email protected]) <http://www.eyrie.org/~eagle/> _______________________________________________ OpenAFS-info mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-info
