[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]

[Cluster-devel] [PATCH 2/2] fence_scsi: always do sg_turs before registration



If a node has been fenced, or its key has been forcibly removed from
device(s) by another node, the victim will see "unit attention" from the
devices if/when it attempts to re-register. This error is normally
reported just once, but will cause registration to fail. We can safely
avoid this by calling sg_turs prior to each attempt to register with a
device.

Resolves: rhbz#640343

Signed-off-by: Ryan O'Hara <rohara redhat com>
---
 fence/agents/scsi/fence_scsi.pl |   18 ++++++++++++++++++
 1 files changed, 18 insertions(+), 0 deletions(-)

diff --git a/fence/agents/scsi/fence_scsi.pl b/fence/agents/scsi/fence_scsi.pl
index 818e1d9..10dc26a 100644
--- a/fence/agents/scsi/fence_scsi.pl
+++ b/fence/agents/scsi/fence_scsi.pl
@@ -46,6 +46,7 @@ sub do_action_on ($@)
 	log_error ("device $dev does not exist") if (! -e $dev);
 	log_error ("device $dev is not a block device") if (! -b $dev);
 
+	do_reset ($dev);
 	do_register_ignore ($node_key, $dev);
 
 	if (!get_reservation_key ($dev)) {
@@ -226,6 +227,23 @@ sub do_preempt_abort ($$$)
     return;
 }
 
+sub do_reset (S)
+{
+    my $self = (caller(0))[3];
+    my ($dev) = @_;
+
+    my $cmd = "sg_turs $dev";
+    my @out = qx { $cmd 2> /dev/null };
+    my $err = ($?>>8);
+
+    ## note that it is not necessarily an error is $err is non-zero,
+    ## so just log the device and status and continue.
+
+    log_debug ("$self (dev=$dev, status=$err)");
+
+    return;
+}
+
 sub key_read ()
 {
     my $self = (caller(0))[3];
-- 
1.7.3.4


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]