Since we still don't have the service plan (we must define a financial source first and then it goes through a couple of layers of lawyers...), my strategy is now to get this fix prepared so that we can pick it up from the support repository for 2009.06 when we get the contract finally (we need a full hand of other fixes therein as well). So I'm absolutely willing to help out as much as I can to find the cause and a stable fix, when NFS v4 will be common this bug will surely bother a lot of people.
We see this problem during fast write access to small files (a couple of jobs on different files on the same filesystem) through the automounter (layout and options as described above). Files sometimes do exist before and sometimes not, so no clue there. I'm not sure wether it is triggered when the mirrormount filesystem is not yet mounted (I'm actually not the one who sends those jobs), but I have the impression that, once it is triggered, it happens often in bursts, like some resource is depleted (NFS queues ?). Jobs afterwards run ok for a while, then the next burst of 'device busy' turns up. What to test next ? Do you already know the source line the EBUSY condition is triggered in bug 6636260 ? How to check if we really hit 6636260 ? Thanks for any help! -- This message posted from opensolaris.org
