MSCS: RPC Server is unavailable during node add

This one seems to be a little specific to my (or other large) environments.

Turns out, despite the fact that my first node and second node could ping each other by short name, fqdn; they could use nblookup to perform a netbios lookup against each other, the second node wouldn't join the cluster.

I checked my network binding order, I checked all sorts of other goodies..

Turns out that because of the large environment and the transitions we're going through, the nodes could resolve each other by their 'real' fqdn, but if you tried a ping, it would show a different dns suffix.

e.g.:

ping nodeb.oldcompany.com

Would succeed, but show:

pinging nodeb.newcompany.com

I added an entry in the hosts file to make sure that the short name of the server was listed with the old fqdn, and that fixed it:

192.168.x.x nodeb.oldcompany.com nodeb

iPhone 3.0 Upgrade Issues....

Only one problem with the iPhone upgrade process to 3.0, and I've encountered it twice now that I've seen:

Make sure that you've got good connectivity to the iTunes store from your PC before you upgrade. It'll save you a lot of trouble. Otherwise you'll end up with an inactive iPhone that will only allow you to make emergency (911) calls until you manage to get it connected. Fun stuff.

Browser wars....

You know, I get awfully tired of the stupid browser wars... even more stupid are applications that are built specifically around one browser or another when there's a whole fleet of standards out there.

Things for my job like this:

-IBM SAN Volume Controller Console: Works under IE, but doesn't look right and it's a pain to use. Firefox for this. Chrome doesn't work well at all.

-IBM 3494 Tape Library: Gives a nice warning when you load the page that you should be using IE to manage it. I haven't had any problems that I've noticed if I use Firefox.

-IBM/Tivoli Storage Manager Administration Center: Well, Firefox has less rendering problems than IE or Chrome, but Chrome is close - and a HECK of a lot faster.

None of these are particularly challenging or fancy interfaces by today's standards... why they have to be so specific to a particular rendering engine I'm not sure at all.

Datacenter racking, power solutions

I have to say, we've been very fond with the hardware we've gotten after choosing SMC to provide server racking for us. They custom build everything (and still have very reasonable choices), and you can work with them to receive engineering drawings of the racks of your choice before you order them. In our case, we picked a nice deep rack with lots of space on the left for zero-u mounting of PDUs.

We also (as they have been partered with SMC) switched to putting in 60A 3-phase power to every rack. We've got an Aphel front-end PDU for each rach, which we then cross-connect with the neighboring rack for the load. (we install all racks in pairs now, side-by-side)

The unfortunate part of this process is actually getting the equipment. We've gone through the process to order nice equipment, then had horrific times actually having it get to our site. 99% of the time, the problem wasn't with SMC, but with the PDU supplier. In the case of our last order, they had been purchased by another company, who had come in and discontinued the slave 15A PDUs we were using... unfortunately, this happened while we were waiting for shipment.

Well, we received the racks right quick. Took us 6 weeks to get our front-end PDUs. We're still waiting for delivery on some slave PDUs that they had to substitute in.

On a good note, the newly substituted slave PDUs will come with custom-made mounting brackets for our rack and instructions on how to install them. (they have a different form-factor than the others we had been getting)

Facebook + Javascript

Simple technote, since I had to look up the solution.

In Firefox (my regular browser), I started getting a message when at the logon page of Facebook that Javascript was not enabled in my browser. This caused issues and make things not work right.

The problem persisted for a couple of days, and I found that IE didn't have the same problem.

Resolution: Cleared my browser cache, forced reload on the page, and we were all good.

Systems Packaging/Shipping

You know, I received a new (well, used unit, new to me) Clariion yesterday from another division in my company. Came with around 10TB usable of 146GB drives in a nice EMC cabinet.

However, the most impressive thing about the whole unit was the packaging. The crate it came in was very nice, easy to open, and had a very nice hinged ramp once it was opened. Took a little wiggling on the crate to get the rack out, but once it started moving at the wheels were all pointed in the right direction, the rack rolled right out. With a couple extra pairs of hands, we had it unboxed in about 5 minutes.

I guess it's not really a technical topic, but it sure did make my life easier. How many times have you received systems that came in and took longer to unbox than to put into production? I think I had some HP blade servers that were like that recently... had to unbox the RAM, extra CPUs, voltage regulators, fiber HBA, etc. What a pain!

Fixing Windows with Linux

For those of you who've ever wondered why it's such a pain to resize the C: drive on a Windows server, here's a nice (and free) solution.

1) Download the latest version of the Knoppix Linux LiveCD and burn it to a CD.
2) Boot it up, open a shell session.
3) 'su' to root.
4) Run 'qtparted' to have a nice graphical tool to resize partitions, including NTFS partitions.

Knoppix is also lovely for giving you read-only access to an NTFS partition when you've destroyed your Windows install. I recently used this method to recover data from a home PC for a family friend where their windows install was completely inoperative. Booted up Knoppix, plugged in a 300GB USB hard drive, and copied all the useful data from the drive before a complete wipe and reload.

Good stuff, good stuff!