OA and VC Communication Issue

[jhani]

We have been running c7000 with different version of VC with single OA for 10+ years. Access to OA and other hardware components thru it has not seen so time critical that we would start using double OA configuration. Also with one OA we have not had any configuration sync problems you might have with two OAs. Having only one OA has not seen any big problem from HPE side on discussions we have had.

We have quite big installation base and therefore we do have spare parts in our premises for almost every component we run. Replacing broken OA has never been problem. It´s easily done from internal spare parts before getting new one from HPE support. Then we have restored OA configuration and re-created OA and VC connectivity.

During last six month we have started facing strange behavior when OA is broken and replaced. After restoring OA configuration VC will ask normally credentials to re-create OA-VC connection. Everything looks normal on VCM side. On OA there is exclamation mark on top on each blade telling that server is not configured for Virtual Connect.

At the first time we had this issue we thought that OA and VC connectivity is not working well and we did reset for OA which did not change situation at all. Then we did reset for VCM. It did go to normal tree of red icons when you do reset of VCM but this time it did not recover in ~15 minutes it usually does. It did not recover at all!

At this stage we open up support case to HPE. After going thru logs they answered that we do have only one OA and therefore connection was lost between components. Only way to recover from this situation would be Power off all blades & interconnect modules (FF) and then remove OA and it´s sleeve. Then place everything back and start again. We did this and problem was solved.

This would be fine if it would be one time problem needed to solved like this. But now it seems that every time we have broken OA we end up on same situation. As said we have run our environment for 10+ years and replaced OA module quite many time without this kind of problems. Firmware version in these problematic cases has been OA 4.40-4.60 and VC 4.45-4.50.

Is there some bug we and support are not aware?

Or has something changed in way components communicate?

Answers we get from HPE support is that this is normal behavior with one OA and you need reseat all components in enclosure after replacing OA. I haven´t seen any document stating that this is way it should work.

[Torsten.]

I cannot see any good reason why not running in a dual OA configuration.

[jhani]

Thanks for reply! Question was not would it work or not in dual OA configuration! It´s more about if we have run big environment of c7000s with one OA for over 10 years without problems why there´s this kind of problems now. It would not make sense to start bying second OA for all the enclosures were currently running.

[sgunelius]

Are you seeing this behavior on all of the enclosures (assuming they're Gen1/2) or isolated to a specific enclosure. If a specific enclosure, maybe the OA tray is going bad and needs to be replaced. I started seeing OA modules fail in our 11+ year old Gen1 enclosures over the past 2 years, received replacement modules from HPE with firmware so old, I couldn't sync them with the remaining module. Even had one fail during a firmware update and needed to be replaced. I'm in the process of replacing 2 of 3 my Gen1 enclosures with Gen3 with VCFF 20/40 G8 modules. Primarily doing this because the Gen1/2 with 1/10 GbE & 4Gb FC VC modules are no longer supported and we've seen odd behavior from BL460c Gen9 periodically losing connectivity. HPE assumes it to be a compatibility issue between the new servers and old enclosures/VC modules. Also, as Torsten stated, why aren't you running redundant OA? I don't think the cost for a redundant OA is that much and considering that supports your management interface to the enclosure, having that redundant is kind of important.