Please don't reply to lustre-devel. Instead, comment in Bugzilla by using the following link: https://bugzilla.lustre.org/show_bug.cgi?id=11147
Created an attachment (id=9466) Please don't reply to lustre-devel. Instead, comment in Bugzilla by using the following link: --> (https://bugzilla.lustre.org/attachment.cgi?id=9466&action=view) patch to remove excessive debug output This patch stops the debug in ptllnd_wait() from being verbose when the timeout is -ve (i.e. forever) and also uses the proper way to specify "infinite" to PtlEQPoll(). The verbose debug was added when we were paranoid about the catamount ptllnd sleeping forever on RedStorm and it shouldn't really have stayed like that. I've not done anything about overflowing timeouts. ptllnd_wait() interprets any -ve timeout as infinite and you could consider a timeout of 1000000000 seconds as damn near infinite, so the fact that the conversion to milliseconds _does_ change it effectively to infinite is a bit of a gnat. However I _do_ think someone should answer the following questions authoritatively... 1. Why is b1_5's __l_wait_event() requesting a timeout of 1000000000 seconds? 2. Why is __l_wait_event() so different in b1_5 and b1_4? _______________________________________________ Lustre-devel mailing list [email protected] https://mail.clusterfs.com/mailman/listinfo/lustre-devel
