Please don't reply to lustre-devel. Instead, comment in Bugzilla by using the
following link:
https://bugzilla.lustre.org/show_bug.cgi?id=11330
What |Removed |Added
----------------------------------------------------------------------------
Attachment #9243|review?([EMAIL PROTECTED]|review-
Flag|m) |
(From update of attachment 9243)
>Index: include/linux/obd_support.h
>===================================================================
>RCS file: /cvsroot/cfs/lustre-core/include/linux/obd_support.h,v
>retrieving revision 1.92.2.8.22.36
>diff -p -u -r1.92.2.8.22.36 obd_support.h
>--- include/linux/obd_support.h 21 Dec 2006 00:35:46 -0000
>1.92.2.8.22.36
>+++ include/linux/obd_support.h 29 Dec 2006 08:18:04 -0000
>@@ -132,6 +132,7 @@ extern int obd_race_state;
>+#define OBD_FAIL_OST_DROP_REQ 0x21c
Note that 0x21c is already in use on b1_5. Please use 0x21d.
>@@ -698,6 +699,26 @@ static int ost_brw_read(struct ptlrpc_re
> if (rc != 0)
> GOTO(out_bulk, rc);
>
>+ /*
>+ * If getting the lock took more time than
>+ * client was willing to wait, drop it. b=11330
>+ */
>+ if (cfs_time_current_sec() > (time_t)obd_timeout ||
Sorry, this doesn't make sense. The current time (i.e. seconds since 1970)
will ALWAYS be > obd_timeout. What you really want is:
if ((cfs_time_current_sec() > req->rq_arrival_time.tv_sec +
obd_timeout) ||
OBD_FAIL_CHECK(OBD_FAIL_OST_DROP_REQ)) {
CERROR("Dropping timed-out read request from %s "
"because taking the lock: "LPX64" took %lu
seconds.\n",
libcfs_id2str(req->rq_peer),
lockh.cookie,
cfs_time_current_sec() - req->rq_arrival_time.tv_sec);
goto out_lock;
}
>+ * If getting the lock took more time than
>+ * client was willing to wait, drop it. b=11330
>+ */
>+ if (cfs_time_current_sec() > (time_t)obd_timeout ||
>+ OBD_FAIL_CHECK(OBD_FAIL_OST_DROP_REQ)) {
Same here.
_______________________________________________
Lustre-devel mailing list
[email protected]
https://mail.clusterfs.com/mailman/listinfo/lustre-devel
