On Wed, Aug 6, 2008 at 1:25 PM, Brian J. Murrell <[EMAIL PROTECTED]> wrote: > dOn Wed, 2008-08-06 at 12:11 -0600, Chris Worley wrote: >> On Wed, Aug 6, 2008 at 11:26 AM, Brian J. Murrell <[EMAIL PROTECTED]> wrote: >> > >> > +rpctrace debug is probably the way to go to see what the client >> > are(n't) doing in terms of keeping the MDS aware of it's existence. >> > >> >> The log from the client perspective starts with a lot of packet mismatches: >> >> 00000100:00020000:3:1217629906.902955:0:5592:0:(events.c:116:reply_in_callback()) >> early reply sized 176, expect 128 >> 00000100:00020000:3:1217633409.730495:0:5590:0:(events.c:116:reply_in_callback()) >> early reply sized 240, expect 128 > > Given the amount of debug provided, this looks like it could be bug > 16534 which is on my plate to test the solution for and land. >
In adding info to bugzilla, you clarified that this is actually bug #16237. This bug is "fixed" and in Lustre 1.6.6... but 1.6.6 isn't released. Is there a work-around for the 1.6.5.1 client (short of rebooting)? Thanks, Chris _______________________________________________ Lustre-discuss mailing list [email protected] http://lists.lustre.org/mailman/listinfo/lustre-discuss
