[Xen-API] XCP bug: SR_BACKEND_FAILURE_58

[George Shuklin <george.shuklin@xxxxxxxxx>]

Good day. 

I think many guys met this message:

Error code: SR_BACKEND_FAILURE_58
Error parameters: , Logical Volume request must come from master, 

...And hopelessly reboot master and get this error again and again. I do
think I found 
1) quick hack to solve problem
2) real bug causing this message

Scenario to reproduce bug:

1) Create pool with shared SR with two hosts.
2) Halt master 
3) Say emegency transition on second host
4) Forget host (older master).

viola: we have single pbd on master to shared storage with
SRmaster:false in device-config. Now we can reboot host, do plug/replug
- nothing will helps.

How to fix bug (in code): when we saying emegency transition we must set
SRmaster=false on older master pbd's  and say SRmaster=true on new
master SR.


How to fix (quick and dirty). You do this for your own risk.

1) Install tidy (via yum or by rpm)
2) do xe pool-list (note pool-master)
3) do pbd-list (note pbd-uuid assosiated with sr and pool-master uuid
from step 2)
4) stop xapi on master (/etc/init.d/xapi stop)
5) cd /var/xapi
6) mv state.db state.db.bak 
7) tidy -xml state.db.bak >state.db
8) vi state.db 
9) Find uuid="YOUR_UUID_FROM_STEP_3"
10) Find SRmaster field nearby
11) change false to true
10) write and exit
11) start xapi
12) check if pbd become SRmaster in device config (xe pbd-list
uuid=FROM_STEP3 params=all)

If not - stops xapi and put back state.db.bak to state.db




_______________________________________________
xen-api mailing list
xen-api@xxxxxxxxxxxxxxxxxxx
http://lists.xensource.com/mailman/listinfo/xen-api