The Subject is wrong - this is a device mapper patch, not an iscsi
one.
--
dm-devel mailing list
dm-devel@redhat.com
https://www.redhat.com/mailman/listinfo/dm-devel
Hi
There are other cases where the kernel could issue or wait on I/O in a
shrinker (for example in inode shrinker). Instead of fixing all these
cases, you should fix the process giscsid to not issue an I/O when it is
not safe.
For example, the lvm and dmeventd processes lock themselves in memo
From: Tahsin Erdogan
If something goes wrong with an iscsi session, the problem is reported
to giscsid via a netlink message. Then, giscsid tries to add a new device
and destroy the old one. During old device destruction, the pending ios
get completed with an error. Without destroying the device