how to trigger ms_Handle_reset on monitor

[agung Laksono <agung.smarts@xxxxxxxxx>]

Hi Ceph user,

I am learning bugs in ceph and am trying to reproduce http://tracker.ceph.com/issues/15113 on my local machine.

On the description, I may see the scenario:

1. ______________________ SafeTimer::timer_thread(), with mon_lock:
2. ______________________ elector: in win_election(), it resend_routed_requests(), and collects the routed requests
3. msgr: in ms_handle_reset(), it reset the session
4. msgr: it waits for the lock
5. ______________________ elector: in win_election(), it handle_command(), but the session is reset, hence it panics.
6. msgr: remove session, and erase related requests from Monitor::routed_requests.

I could understand the whole flow of the bug except ms_handle_reset(). What's a thing that triggers this ms_handle_reset()? 

I could not see this function is called in any part of the monitor class. I've tried to reproduce this on a cluster with 3 MONs and 4 OSDs 

and put the printf whenever this function's called. However, It seems arbitrary. Perhaps anyone can help me to explain this? 

thanks

--

Cheers,

Agung Laksono

_______________________________________________
ceph-users mailing list
ceph-users@xxxxxxxxxxxxxx
http://lists.ceph.com/listinfo.cgi/ceph-users-ceph.com