RAID Monitoring Options

[Rob Boudrie]

I recently rented a Dual Xeon server w/SCSI RAID 5 (3 drives). When I got it, I was told that ThePlanet monitored the RAID for drive failures. I filed a ticket asking (figureing the techs would know more than sales) and was informed they do not monitor it, but would attend to a RAID alarm if they noticed while walkking by the server. The suggested I use afacli to monitor the RAID array.

Do you have any documentation on how to monitor with AFACLI or, better yet, a monitoring script for a cron job? RAID 5 isn't that useful if I don't learn about a failed drive promptly so I can have it replaced before there is a second failure.

[jbyers]

I tend to recommend using this script from Dell's linux website

http://linux.dell.com/files/aacraid/aacrai...ring_script.txt

Here's the actual documention that comes with the Dell PERC RAID controllers:

http://support.dell.com/docs/storage/57kgr/cli/en/index.htm

[Rob Boudrie]

I already found that script and downloaded afacli, however, from what I can see afacli may not be the correct program for the Adaptec controller in my box. I see a "megaraid" device but no "afa" device in /proc/devices. Adaptec has raid management tools on its website, but you need the serial number of the controller to download them.

I have a ticket in to ThePlanet and I expect I'll have the info shortly after the weekend is over.

I'm a bit confused about the mixed messages regarding RAID - sales told me that ThePlanet would monitor raid; the ticket support people told me they would respond to a raid alarm if they happened to notice, but do not do any monitoring. Once again, I expect that will be straightened out shortly as well.

Planet folks - how about a "RAID user's guide" for buyers who order this feature?

[James Erickson]

For the most part, we don't monitor the raid array unless it alarms in the datacenter, then it will be picked up on the patrol that goes by once every few hours. Can you PM me your ticket number and I'll see what I can do to get it updated for you.

[Rob Boudrie]

ThePlanet did a great job of assisting once it got to the correct tech.

However....

ThePlanet's handling of RAID is seriously lacking.

1. Customers are not told RAID is unmonitored or given tools to monitor it (unless they ask).

2. Customers are given incorrect information by sales. Sales told me that "we will monitor the raid for you"

3. Tech support, in my case at least, gave out factually incorrect information telling me that AFACLI was the CLI for my raid controller even though that is not compatible with the Adaptec controller that they (in the same email) told me I had.

4. Another tech support person installed "megarc" for me and I now have access to the controller. All I need to do now is figure out how to set up a cron job to monitor the RAID status so I will find out if a drive goes bad.

RAID5 is not all that useful if there is not a mechanism to detect and replace bad drives (which is why really high end RAID systems actually submit their own service calls when a drive goes bad).

[Jeff]

I haven't dealt with sales in a while now, but I agree with you (in general, for every company) that it's very important that all the sales people know the systems they are selling inside and out. A good sales person should be curious about what type of raid card is in the server and how it works and be able to tell the customer or research it by asking their coworker and then get the customer the right information. Sales people that don't know the product they're selling are a big turnoff - a remember a couple years ago I got someone at another company who couldn't even tell me the brand of motherboard they were using, so I moved on; no matter how much per hour that guy was being paid it was too much since I bet he turned away more long-term business than he signed up.

RAID5 is not all that useful if there is not a mechanism to detect and replace bad drives (which is why really high end RAID systems actually submit their own service calls when a drive goes bad).

It sounds like all the raid controllers have audible alarms, so when a drive fails a physical buzzer goes off that is then detected when they walk through the datacenter every few hours. So you have a few hours more risk than if you had software monitoring, but with patrols every few hours your odds would have to be very bad to have a second drive in the array fail in those few hours.

[eth00]

It sounds like all the raid controllers have audible alarms, so when a drive fails a physical buzzer goes off that is then detected when they walk through the datacenter every few hours. So you have a few hours more risk than if you had software monitoring, but with patrols every few hours your odds would have to be very bad to have a second drive in the array fail in those few hours.

I don't know what cards do or do not have a buzzer but I know that not ALL do. I have had drives fail and even though it may be failed for a few days EV1/TP never contacted us.

On the flip side we have also had a ticket created because the machine was making some beeping.

I don't know the official word on if all/some have alarms, maybe he times we have had it happen were anomalies.

[Rob Boudrie]

but with patrols every few hours your odds would have to be very bad to have a second drive in the array fail in those few hours

It was basically an "if we notice", not "we patrol for raid alarms". I suspect a "patrol" is just someone walking through the room looking for signs of obvious physical problems, not someone watchinzeeblinkinzeelights.

[Rob Boudrie]

One of The Planet techs located a script that drives megarc and can be easily modified to do auto notifications - fantastic and competent help (just took a few iterations to get the ticket referred to the right person).

They should provide this as "basic info" whenever someone signs up for RAID.

[markcausa]

Yeah, The Planet definitely needs a knowledge base incorporated into their forums or something...

PhireFast at one time had a whole sepetate CMS on a subdomain as our knowledge base, then we started using a sub forum on our IPB for that.

Now that we have Kayako, it's all there too.

Maybe some type of blog-like system for knowledge can be put into Orbit?

[Chris Interrante]

Agreed. I am working on getting a notice out to customers to assist them with monitoring the health of their RAID arrays. I have noticed that this is an area of opportunity for The Planet. Thanks for all the posts on this topic!

[XGhozt]

Yeah, The Planet definitely needs a knowledge base incorporated into their forums or something...

PhireFast at one time had a whole sepetate CMS on a subdomain as our knowledge base, then we started using a sub forum on our IPB for that.

Now that we have Kayako, it's all there too.

Maybe some type of blog-like system for knowledge can be put into Orbit?

Wasn't there a huge knowedgebase a long time ago at TP? I've got tickets linking me to some of the articles.

[James Erickson]

You may be refering to this one:
http://support.theplanet.com/knowledgebase/users/search.php

[Creed3020]

You may be refering to this one:
http://support.theplanet.com/knowledgebase/users/search.php

Where has that been hiding. I couldn't find an external link to that anywhere on the TP website.

Going to paruse that now

[ajz4221]

You may be refering to this one:
http://support.theplanet.com/knowledgebase/users/search.php

They started that a while back but the link "disappeared" one day when everything started changing fast and often (mainly the site design).
(It may still be in orbit and I just havn't seen it...)
And of course, I didn't save a link back then.

[alden]

I just ran across this thread, as I was curious as to how my RAID array is being monitored. I thought it might be nice to have a belt-and-suspenders approach to monitoring, in that I would be notified along with TP techs when a drive failed.

I submitted a ticket and found that TP still does "monitoring by walking around"! I was told that I could "install a third-party monitoring tool". But of course, I don't know what my hw raid controller is! I'm willing to dig a little, but I don't know where to start.

A kick in the right direction would be appreciated!

[Tomy Durden]

One of the DC Ops techs in each data center does a walk of the floor every four hours. On Dells, a failed array will usually trigger an audible and visual alert. Once noticed, a ticket is created for the visual/audible alert in which we request to investigate.
Most of our customers on Dell servers use Dell's OpenManage to monitor the RAID. You can set it to pop off an email upon a degradation. IPAlert doesn't have the capability of monitoring RAID status, or at least I haven't found it yet.

[Austin P]

Most of our customers on Dell servers use Dell's OpenManage to monitor the RAID. You can set it to pop off an email upon a degradation. IPAlert doesn't have the capability of monitoring RAID status, or at least I haven't found it yet.

Anything that will support 3-4 Drives is going to be a Dell Chassis so you can bet it will make a loud alarm that will be caught on prem check. As I understand it, we will install open manage for free for the customer.

[CyberSEAL]

Under "Hardware Details" in Orbit I can see I have the following RAID card:

Adaptec \ 4-channel SAS/SATA RAID PCI Express \ 3405

I have not yet been able to locate monitoring software for it. I have an inquiry in w/ Adaptec...in the meantime, was wondering if anyone on here has found a solution?

[CyberSEAL]

UPDATED:

I'm using Adaptec Storage Manager's CLI tool, included w/ their release for Linux...woot:

http://www.adaptec.com/en-US/downloads/sto...SI+RAID+2130SLP