Bladewatch.com

Hi Martin

Sorry to ask a quite possibly stupid question, but is there any impact or result of clearing down the Integrated Management Log, my IT guy was telling me I should do this as we’re trying to fix a problem with memory on our DL360 G3.

Thanks

Fredrick

My response:

Hi Fredrick,

There’s never a stupid question, I’m happy to help. The IML (Integrated Management Log) notes hardware events maintenance events and component failures, like memory alerts or disk failures. I have cleared down the IML logs before, it’s completely safe to do and is sometimes one of the first things you do to check if the error has been fixed and so that the log does not get read incorrectly.

For example sometimes in the past when we have received a memory alert on a Proliant server, I have cleared the log, powered down the server, reseated the memory chip to see if the error happens again before running the diagnostics or swapping the memory. Hope that helps, if you’ve got any other questions, do give me a call.

Regards

Martin

I wrote this post and screen shot about the IML, check it out for more infomation:

http://www.bladewatch.com/2007/03/02/whats-the-hp-integrated-logiml/

http://www.findmyfirmware.com/firmware/IML.JPG

The IML logs all hardware type events the machine has experienced, you can clear it manually or by script, it will have no effect on the operating system.

It’s job is to monitor the box through the agents and note any issues, like bad memory chips, or blue screens/asr errors etc.

One of the things that we went over at the HP Bloggers event was the support plans for the HP C-Class enclosure, they have extended support for the existing blade enclosure until 2016. Enclosure support has been a topic of conversation for many customers, the concept being that they seek avoiding investing in an enclosure if it is set to be replaced our unable to support future technologies. It’s great to hear, you can find out more here. The same question is posted here.

I put this together for a client when they asked what to do if their server became disconnected from the network, and what their networks guys would typically ask in order to fix it.

1. Note the server network card speed and duplex – if they are wrong, this can cause problems – control panel, network settings view the card hardware settings speed/duplex tab
2. Check the server ip address, subnet mask and default gateway – if these values are incorrect you might not be able to ping out or receive traffic – control panel, network settings, ip configuration
3. Write down the mac address  – ipconfig /all should give you this or if you click the network card in control panel/networks it should be there as well (or if you have a network card management utility like the Broadcom one)
4. Get the patching team to trace the network cable to the switch to confirm the network route and patching are clear, also have the networks team check for error disabled or crc type errors.
5. Have the patching team run a fluke across the cable to check if they can connect using the same values, ruling out a patching/network issue.
6. Try configuring a laptop or even the second network port with the same details to see if it is a network port issue on the card.

As with anything though it’s all about the context, the first thing to consider is to do a health check of the server, at the same time have the networks team check the network port for errors and then have the patch cable checked. It can be a number of issues at fault, but be careful to note what you are changing in case you need to role back, and if in doubt don’t be afraid to ask for help.

We were having a discussion over dinner the other day about the concept of hot swap. It all started as a result of a colleague asking if I had ever swapped one of the internal hot swap fans on a server before without taking the server down. The answer yes, but I did highlight that there are degrees to doing this, although the fan is hot swap and there was no disruption that does not mean it is something that can be done without following the change process which resulted in a lengthy debate.

Regardless in terms of best practice, despite some of my colleagues opinions, if you can, check swapping the hot swap components to be sure particularly the more you delve into the system, a power supply or hard disk is fine, but once you get into fans and memory dimms I tend to err on the side of caution. A little bit of paper work and notification is best practice and can help avoid issues should anything go wrong:

“We’re going to swap this component, this can be done online and without impact, we have carried out this activity before and thought we would notify you.”

Not hiding the risk, but at the same time not advertising the chances that the world might end for the sake of approval, trust and re-assurance.

I got asked by a colleague the other day about upgrading firmware on servers and whether it takes the server off the network”

“is it something I can do online (below the radar), or does the server disconnect from the network?”.

In answer to her question I responded that the network card will briefly disconnect from the network but you might not necessarily from a user standpoint see an outage, regardless it is something that you probably want to do out of normal hours or with user approval just in case. It doesn’t always disconnect from the network and even if it does it might only drop a ping or two, but for the sake of goodwill and best practice, you’re probably best letting the users know that there may be a brief outage.

I got asked the quickest and easiest way to tell what version of the Compaq or HP Proliant Support Pack (drivers and utilities) are installed.  There are two ways which are fairly straight forward, of course you could check the registry and run the new PSP to tell you, but the two ways I would use are:

Control Panel

Double click the Compaq Management Agents (or HP Management Agents), it should tell you the version installed for example 6.10

The other way is to log on to the server web agents which are installed with the support pack to monitor the server health and manage it remotely.

This can be accessed by typing http://yourservername:2301 and logging in using either the password you have set, or if it’s a newer version of the agents an account which is in the local administrators account. So for Janet’s fileserver files1 it would be http://files1:2301

I got a quick question the other day from a reader, Janet she was thinking about upgrading the drivers on their file server running Windows 2000.

I’ve got a DL380 G2 which we use as the file server in our estate agents, it’s working fine, but the IT guy said that we should upgrade the drivers and I wondered if this is something I could do and if there was anything that I should know? I’ve managed to search and I know it’s running the Compaq Proliant Support Pack 5.5?

Thanks

Janet

My response:

Hey Janet,

Great to hear from you and glad to hear that your server is doing ok. Upgrading the Proliant Support Pack is easy, for your particular server it’s here. I usually download the psp-7.60.w2k.i386.exe (82 MB) file to my pc and then copy it to a place on the server like c:\temp. Double click the file and it will ask about extracting it which might take a few minutes. You then run the file setup.exe, a menu system will appear, if I remember correctly, it’s next, then the first bundle tab, then select all, and then install.

Issues to worry about, if you are upgrading a really old version of the drivers you can sometimes have a driver/utility problem but that is quite rare, also when the network card installs, the server can fall off the network whilst it reloads the driver and the network utility, but again that’s nothing to worry about. In terms of risk, it’s not particularly difficult and carries a relatively low level of risk, if in doubt speak with HP or your service provider.

One thing to note is that if you are upgrading from the Compaq Proliant Support Pack to for example the one above 7.60, the services name will change from Compaq to HP, but that is only a problem if you have monitoring tools looking for those services and does not affect functionality.

It should take about 10 -20 mins in total.

Regards

Martin

http://www.bladewatch.com/about/findmyfirmware/

I’ve just added some of the HP Proliant Blade servers firmware and product information to the iphone application FindmyFirmware, so if you’ve downloaded it, the number of servers should increase and now include the HP Blade servers, I’ll be adding the new HP Blades later tomorrow, all you need to do is connect your iphone to the web and open the application, it will download the new file from the Bladewatch server.

http://www.hp.com/hpinfo/newsroom/press/2011/110119xa.html

HP and Microsoft Corp. today announced a portfolio of four new converged application appliances that fuse applications, infrastructure and productivity tools into a single system. These solutions help organizations optimize employee productivity and decision-making, while simplifying the delivery of applications for IT.⁽¹⁾ The HP Business Decision Appliance is available now to run business intelligence services, and the HP E5000 Messaging System for Microsoft Exchange Server is planned to be available in 45 days to run messaging services. Two other appliances are expected to be available later this year.

Until now there have been two ways to deploy critical business applications: highly customized deployments that take too long, or proprietary stacks of applications and infrastructure that are inherently rigid and slow to change. The result is only 32 percent of IT projects that deliver critical business applications are rated as “successful” by the organizations implementing them.⁽²⁾

This does sound cool, anything that HP and Microsoft can do to help meet business needs and improve organizations application delivery or decision making has to be a good thing, I’m off to read up more about the announcement.

I got asked the other days by one of my Project Manager friends (Mike) about the technical issues involving allocating storage to a server. His viewpoint, he was being quoted between one and two man days to allocate some storage to a server. This of course was billed to his project and he was not best pleased about it.

Allocating a specified time value to a task is always something that can be taken out of context. Swapping a disk on a server takes five minutes, but if the disk fails to rebuild and the server crashes, the activity or even the outage could be significantly longer.

The length of time to cut/allocate storage will depend on the complexity of the request, 100GB on any block or meta size, combined from any array of disks without concern about alignment or performance might take an hour or so (even less dependent on the automation). Also, the availability of the storage in terms of process – some organizations do not allow you to make changes to the existing storage platforms to avoid configuration or performance issues.

To me, the right kind of ‘man day’ effort would be a day, but then I am not a storage person, I’m not looking after Mike’s storage requirements, so with that, let us understand that the man day thing is more a cost indication a vehicle for getting the project to pay towards its on boarding costs, and that any cost would ideally be avoided, but we’re not in that space, so next..