[Bugs] [Bug 1181048] lockless lookup cause disk to be kicked out

[bugzilla at redhat.com]

https://bugzilla.redhat.com/show_bug.cgi?id=1181048



--- Comment #21 from lidi <lidi at perabytes.com> ---
I have a hypothesis that described below.

1. There's a disperse volume with 3 disks(A,B and C).

2. On first session a write is executing, all data was writen, then it would
update size and version 

3. At this time, on second session a lockless lookup was executed first, and
its return value indicated that disk B was broken, disk B would be marked as
broken in inode or fd. 

4. Before the self-heal was executed, disk C was killed or disconnected, so
nothing would be healed.

5. Then update size and version, the operation would dispatch to disk A and
C(ec->xl_up not be updated yet), so finally only disk A was updated.

I just don't know how to verify this. Do you think it could happen?

-- 
You are receiving this mail because:
You are on the CC list for the bug.
Unsubscribe from this bug https://bugzilla.redhat.com/token.cgi?t=aLUXWDPf2U&a=cc_unsubscribe