Unsolved
This post is more than 5 years old
1 Message
0
469
June 26th, 2008 00:00
problem of LCC link failover
Hi,
I had setup a CX3-10c with 2 host runging oracle RAC 10g.
basically, everything is fine and going to the
UAT status, the failover test on the fibre cable to san switch and san switch to hosts failover test are OK, but the LCC connection's failover test is fail and the host will restart once the unplug a LCC cable from SPA or SPB, it cannot takeover the disk access of the disk which owned by the opposite processor.
the connection of the existing system is SPA BE0-> PRI(A), SPB BE0->PRI(B).
disconnect either one of the LCC cable the system fail.
any suggestion to solve that problem?
thanks alot.
Rick
I had setup a CX3-10c with 2 host runging oracle RAC 10g.
basically, everything is fine and going to the
UAT status, the failover test on the fibre cable to san switch and san switch to hosts failover test are OK, but the LCC connection's failover test is fail and the host will restart once the unplug a LCC cable from SPA or SPB, it cannot takeover the disk access of the disk which owned by the opposite processor.
the connection of the existing system is SPA BE0-> PRI(A), SPB BE0->PRI(B).
disconnect either one of the LCC cable the system fail.
any suggestion to solve that problem?
thanks alot.
Rick
0 events found
No Events found!
nandas
6 Operator
•
1.5K Posts
0
June 26th, 2008 08:00
Welcome to EMC Support forums. I did not see any reply to your question - thus thought of putting my 2 cents.
As per my understanding you can not simply unplug the LCC cable to complete a LCC failover test. The backend connectivity is provided by the two LCCs which are redundant to each other. So, taking out cable from one may not initiate a LUN trespass.
I guess, when you take out the LCC cable - all LUNs are available and owned by their designated SP - right?
I'll not encourage to run this test of taking out LCC cable - as that is the integral part of the backend bus connectivity. I 'll trust the EMC architecture that, if any LCC fails, the system will work.
Other experts may correct me or provide valuable suggestions/comments.
Thanks,
Sandip
RRR
6 Operator
•
5.7K Posts
0
June 27th, 2008 01:00
It's a fact: if an LCC fails, the systems keeps working, the management stations dials out and your CE will contact you about this ands you'll have a new LCC in a few hours (except when you're in Bagdad or something).
DGM3
238 Posts
0
July 8th, 2008 23:00
Can you please advise what rev of Flare code is running on the array?
For most configurations, the failure of an LCC will result in an error being returned from the owning SP to the server, and host failover software will be responsible for re-routing the IO to the peer SP. If the failover is not occurring, checking to see if host failover software is configured properly would be advised.
That said, Release 26 of Flare does include an enhancement where an SP can now access drives via its peer SP - an SP can essentially now internally re-route around an LCC failure / disconnection.
Thus it would be important to know if the CX3-10c is running R24 (in which case an LCC removal/disconnection will generally result in host failover software being invoked) or R26 (in which case the array can now internally re-route around an LCC removal/disconnection).
Regards,
DGM