Hot swap Perc 5i RAID controller battery in Dell Poweredge 2900

dlethe

"If it doesn't specifically mention something will work, assume it won't"

Always good advise.  That is the direction I am heading.  I am currently planning on taking the environment offline.  

However, I put the question to EE to see if this was not necessary.

Thanks for the input though.

karislove

I'm intimately familiar with the firmware as this sort of thing is what we do.  As I think about it, it probably won't "hurt", provided you take static electricity seriously .. but it would never work properly.  The controller has to reboot to allocate the memory. If battery is dead, then the write cache will be disabled and read cache/buffers are increased internally.

So even if you could connect it, you not only won't benefit until a power cycle, but you put your data at risk for doing so.  It is possible the controller would walk all over live data.

dlethe,

Thanks for waving me off of the idea.  I think I will take the safe road and continue as planned.  Management will have to adjust.  Since I have you on the line, do you know of any gotchas when replacing the PERC 5i battery while offline?  I have never done one before.

karislove

Just make sure that you aren't in a degraded condition, or are rebuilding, or anything that means the controller is in stress.  If need be, go through a full power-off cycle to the BIOS and make sure it is healthy first (no need to boot the OS, just power down gracefully, wait 30 secs to clean out capacitance, boot to BIOS and make sure the controller says everything is fine.

Then shut down, replace battery.  Power up, detect.  It generally has to go through a reconditioning cycle before the write cache gets turned back on, so may be 24 hours or so before you see improved write speed. -- I assume old battery is dead, not just weak.

dlethe,

Actually the battery is reporting all good at the moment.  However, whenever the battery goes into a learning state whether scheulded or manual the log starts filling up with errors.  If the battery is reporting good when I power off will that effect the replacement?

karislove

A learning state?  You mean reconditioning state.  Reconditioning is normal, and the event log entries should just be warning entries that can be ignored.

What, exactly are the messages?

dlethe,

I believe the activity is reconditioning but the Dell Open Management application calls it a "learn cycle".  It is where the controller disconnects the battery from the onboard charger to drain it and recharge it to verify it's integrity.

I don't have the exact error to give you at the moment as i am not in front of the machine.  However, (from memory) the error indicated the battery failed some sort of test.  The error log filled up with these "red x" messages untill it was to full to accept any more messages.

I will see if I can log in later and take a look.

Thanks,

karislove

Go ahead and replace it then, just looking out for you in case you thought reconditioning cycle was an error not a warning.   You can't argue with red X messages and failed tests.

dlethe,

Sounds like a plan.  Thanks for sharing your expertise.

karislove

I had my Perc5/i battery fail -  I tried to hot swap it.   The only thing that appeared in the log was a message saying that the battery was removed.   Nothing was logged saying that the new battery was connected.   As far as the controller is concerned, there is no battery.  

Even so, this is useful because I've replaced the battery, now when I'm home tonight, I can probably reboot the server and the battery will (should) appear again..   and I don't have to stay at work late for the replace/reboot.