A common question I get when talking with others about our Cisco UCS production environment is if we have had any issues that required us to deal with Cisco TAC. Like with anything we have had a few things that required a call. By the way, the phone number is the same for any Cisco product. Here are a few examples for you.
One of our first calls had to do with a failed 8 GB DIMM in one B200 M1 server blade. We noticed a warning light on the blade and went to the UCS Manager to investigate. We were able to quickly drill down to the effected blade’s inventory and go to the memory tab. This screen provided the details of the failed DIMM’s slot location and confirmed it’s failed status. Since the workload running on this blade was VMware ESX we put it into maintenance mode, powered down the blade and replaced the DIMM with a spare. It was time to open a ticket with TAC.
The TAC engineer took down our information and sent out a replacement DIMM within 4 hours and we were done with the ticket. I asked our server person what he thought of dealing with TAC and he did not expect it to be that easy. Typically in the past with other server vendors we would have had to run a diagnostic tool to determine which DIMM and then open a trouble ticket. We would have to down the server, re-seat the DIMM, and wait for it to fail again. Once it failed again then we would get a replacement. So this call process with Cisco seemed to be smoother.
Another trouble ticket was related to a VMware ESX host that, post a reboot, would not see the boot partition. After some troubleshooting, it clearly was a ESX OS issue and our VMware admin was ready to re-image the server. However, we thought this would be a good test for Cisco TAC so we opened a ticket. We were surprised when TAC gave the case to an ESX server person at Cisco who within 20 minutes had resolved the issue and the server was back in production. So our expectations were exceeded again.
The one trouble ticket that took sometime was when we wanted to install Windows 2003 standard 64 bit bare metal on a blade with the Emulex interface card. This is easy to do with Windows 2008, however the challenge was getting the right drivers on a media type that the Windows 2003 installation process could recognize. It wanted to see the drivers on a either a CD or floppy disk which you provided by emulating the media. I personally did not work this ticket but it took time over 3 days to get everything completed. In the end we now have a process down and 2 servers in production.
Overall, Cisco has exceeded our expectations when it comes to dealing with trouble tickets around the UCS products successfully. It has been clear to us that Cisco has put the resources into support and have the right folks in place to deal with a variety of potential issues customers may run into.