On Mon, Mar 13, 2017 at 12:01:45PM -0400, J. Bruce Fields wrote: > That wait doesn't sound bad at all. > > The server could track round trip times to the clients holding > delegations and use that to estimate the right wait time, but hopefully > that's overkill. > > I think the occasional retransmitted truncate probably isn't a big deal. > Seems like the kind of thing the reply cache would handle well. I haven't thought about what it'd take to implement. Just for testing purposes, maybe stick a 50ms wait and a retry in the JUKEBOX case in anything under fs/nfsd/nfs3proc.c that looks like it could break a delegation. Jeff's proposal should also be pretty easy to prototype. --b. -- To unsubscribe from this list: send the line "unsubscribe linux-nfs" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
