ProLiant Servers (ML,DL,SL)

Unusual number of FBWC modules failing... anyone else with this experience?

 
PDP-Fan
Valued Contributor

Unusual number of FBWC modules failing... anyone else with this experience?

Since a couple of months we encounter an unusual number of failed FBWC modules. Some have problems charging the caps while others have "data in cache" that cannot be written to the drives.

We tried all kind of tricks to get these modules working correctly (changing caps, trying different setups) but at the end we had to replace them all.

 

At the moment I have the feeling that , while the caps themself are flawless, these modules fail more often than the batteries of the older BBWC's. Only the modules are more expensive....

 

(for example PN: 505908-001, 1GB FBWC, DL380-G7)

 

Does someone experience the same?

Does someone know a trick to fix these modules?

 

 

Chris

 

 

***********************************************
"If it seems illogical... you just don't have enough information"
2 REPLIES 2
waaronb
Respected Contributor

Re: Unusual number of FBWC modules failing... anyone else with this experience?

I have one DL380p Gen8 right now, only in use for about 4 months, where the FBWC cap isn't charging.

I can power the server down entirely and let it sit a while, and when I boot up it's fine. But then after 24 hours or so, it pops up the warning saying the super-cap isn't charged and write-caching is temporarily disabled.

If I power down again and let it sit for a while and power up again, it repeats the process... it's fine for a while, then discharges enough to get that error.

This is on the P420i that it came with. That server also has an extra P420 controller with it's own super-cap that runs the other 8 drives in the expansion cage, and that one hasn't had a problem at all.

I wonder if there's a loose connection perhaps, or if it's just a bad cap, but since it's in a remote colocation, it'll be a while before I can check on it and it's not a big enough deal to warrant a site visit.

Since it's in a managed space with redundant power, I can set it to do write caching anyway, even with the cap unavailable. There's very little chance that my redundant power feeds to our cabinet would both go down since this is in an Internap datacenter...I mean, short of Hurricane Sandy that is (it did happen then... clogged diesel lines knocked the generators out, and their UPS only ran things for another 30-60 minutes without utility or diesel power). :)
PDP-Fan
Valued Contributor

Re: Unusual number of FBWC modules failing... anyone else with this experience?

This sounds like one of the problems we had too.

I think there is a timeout counter that resets when the server is off. Then it waits some hours for the cap to reach the fully charged voltage. If that is not achieved within the time, it goes to an error state and disables the cache.

 

There are 2 possibilities... either the voltage is not reached... or the sensor that checks the voltage measures a wrong value.

 

However you have not many possibilities to fix that. Just replace the cache module. To be on the safe side replace the capacitor too. Although I never had a faulty cap.

***********************************************
"If it seems illogical... you just don't have enough information"