General
cancel
Showing results for 
Search instead for 
Did you mean: 

Degradation in performance before disk array accelerator battery charge error reported

jhodder
Occasional Contributor

Degradation in performance before disk array accelerator battery charge error reported

Hi 

recently we had a HP ProLiant DL360 G5 on CentOS 4.9 that ran into a "POST Error: 1794-Drive Array - Array Accelerator Battery Charge Low".  Looking at performance stats before the error I can see that IO wait was peaking at 12% where as now its at 3% and can also see increased disk utilisation before the issue too.   

 

Is there any explanation of why an issue with the battery charge leading up to the failure might have caused the increased load, disk util etc?

 

Does anyone know the specific behaviour of the write cache based on the status of the battery e.g. charge <75% then no write cache, or is it more comlpex than that? 

 

Thanks 

 

Jason

2 REPLIES
Steven E. Protter
Exalted Contributor

Re: Degradation in performance before disk array accelerator battery charge error reported

Shalom,

 

Battery being low can lead to erratic performance on the RAID card. It is very important to act on these warnings and replace that battery when the warning is received.

 

SEP

Steven E Protter
Owner of ISN Corporation
http://isnamerica.com
http://hpuxconsulting.com
Sponsor: http://hpux.ws
Twitter: http://twitter.com/hpuxlinux
Founder http://newdatacloud.com
jhodder
Occasional Contributor

Re: Degradation in performance before disk array accelerator battery charge error reported

Hi Steven, 

 

thank you very much for your feedback.  So I understand correctly are you saying a battery with low charge but still above the 75% threshold (ie 100-76%) can still affect performance?

 

This is the key fact that I am trying to understand.  We definitely saw issues on the servers when the charge dipped below 75%.  But before the event we so no errors and exrta load on the box.  That extra load didnt cause a performance issue but it was noticeable and so far unexplained.  It could just be down to our application (in this case tomcat) slowly eating up resource over time but I am also out of interest pursuing the possibility of the battery being a contributing factor.

 

Thanks again

 

Jason