28 Jan 2008

On Updates and PERCs

Author: q | Filed under: Frustrations, Tools

I”ve had a few comments show up on the series of PERC 5/i posts I had early in 2007. There have been a few questions about the status of things, so rather than respond in the comments, I thought I”d summarize what I know at this point in a separate post.

Bottom line, the alarm status of the PERC 5/i has not changed, nor will it in all likelihood. In discussions with the engineering folks at Dell, apparently there were a number of people who complained about the alarm with such ferocity that the design team decided it was best to take it out altogether. Akin to driving in a nail with a sledgehammer, I think this was a misguided and completely incorrect overreaction to the problem. If Dell wanted to make the default setting on the controller to have the alarm OFF instead of ON, I would have been fine with it. Make me enable the audible alarm if I want it, but leave it off for those who don”t. But to completely remove the functionality is just beyond me.

At this time, I cannot tell if the next series of the PERC controller will have this functionality restored. The engineering folks that I spoke with said that the feedback they”re getting is still in favor of having the alarm removed. I said “you”re talking with the wrong people, then.” I invited them to have their researchers include me in their feedback request, and I”d be more than happy to add the logical reasons to have the ability to have an alarm present but quiet by default. If you”re in the same boat as I am,please take a moment to contact Dell Support and voice your thoughts on the matter. Apparently (as I”ve been told),they listen to loud feedback.

Secondly, on the issue of proactive monitoring, Dell still does not have a tool that will generate an alert if the array goes into a degraded condition. We have been using HoundDog to provide proactive monitoring of the health of the array for my systems with the PERC 5 controllers, and it has worked very well for our operation. You do have to install the Server Administrator tools to generate the SNMP alerts that HoundDog picks up on, but cost of the HoundDog service and running Server Administrator on the box is far less than the cost of not knowing that an array is having trouble. There are, of course, a couple of quirks. One, the SMNP trap will alert on battery conditions on the controller as well, and not tell you it”s a battery issue in the alert. If the server gets powered off for any reason, the battery on the controller starts “draining” to maintain the configuration information, and when the server is powered back on, the battery goes into “recharge” mode, which triggers an SNMP even in Server Administrator, and then by HoundDog. I haven”t been able to find out how to modify that, but it”s probably good to know about battery conditions, so I”m leaving it alone. Two, not all Dell servers run Server Administrator, specifically the SC-series servers. I have one SC server with a mute PERC controller, and I”m tring to figure out how to monitor that. SNMP doesn”t work, as Server Administrator isn”t present to generate the SNMP configuration that HoundDog is looking for. I downloaded and installed the LSI software, but it has no mechanism for generating alerts, so I”m still digging on that one.

So we”re not fully there yet, but getting comfortable. I will not be purchasing an SC-class server from Dell in the future, but outside of that, HoundDog has given me what I need to keep my team alerted to any problems with array controllers at a very minimal cost. I would still prefer to have the option of dealing wiht an audible alarm, and I still mention my frustration with that every time I call Dell support (which really isn”t that often, but I did speak with the original tech who took my call last February over the past week about another issue, and he rememberd who I was and brought up the subject himself).

Leave a Reply