I have been trying to figure out how to get our Bladecenter blades to talk properly to the SAN controller. The SAN has two fibre-channel controllers, with two separate fibre paths to the data. I created a volume under SLES9 on one of the arrays, and put some data on it. When i transfer it from one SAN controller to the other, it moves back to the first controller when I access the disks under SLES9. This confuses me, and tells me that I may not have the multipathing driver for the Qlogic controller cards set up properly. Hmm..
We’ve got the IBM Bladecenter management stuff all configured. We had started out using the lab IP address scheme, and just selecting a bunch of addresses for the Bladecenter components, but then this morning on the way in to work, Stuart and I both had the same thought: Since we’re setting up all this stuff in the Bladecenter with the intent of rolling everything into production, we should be configuring it with the final IP address scheme instead of a temporary one. We discussed it and decided that it would be easier to reconfigure the lab to conform to the eventual IP scheme of the Bladecenter than the other way around.
Here’s me installing stuff on the Bladecenter from my iPod.
I spent this morning reconfiguring the lab firewall (which I buggered up and had to reconfigure), my two lab VMware servers, and my lab VMs and workstations. We also have one blade configured as a starting point in the Bladecenter to be used as an installation server for the rest of the blades. It’s coming along.
Today we get to celebrate the fact that we no longer have a Liberal government in Canada. The Conservative party has won the federal election, and will make up the next government. Unfortunately, it’s a Conservative minority government, which means the Liberals will be able to impede the work that the Conservatives try to do.
The biggest upside is that we’ve gotten rid of a morally bankrupt party as our government. The Liberals in power were corrupt and excelled only at making things financially sweet for themselves and their cronies. The wonder of the election to me is that many seats in the house of commons were retained by Liberal party members, especially in the big centres of Montreal, Toronto and Vancouver. Re-electing Liberal candidates after that corrupt, thieving bunch of criminals screwed us all over for the last 12 years was like saying “Thank you sir, may I have another” after being slapped in the face.
At least a lot of Westerners are feeling like their votes counted for something for the first time in a long time. I know I feel that way.
James and I assembled all the blades, adding memory, drives, cpus and fiber-channel controllers, on Wednesday. Yesterday James and Stuart assembled the storage arrays and hooked them up. This morning our 220 VAC power distribution units arrived and we were able to power up the BladeCenter. Yay!
Stuart booted each blade and updated it’s firmware, and I booted each blade and created a RAID0 array on each blade’s pair of on-board SCSI disks. We will most likely be booting from the on-board disks, and storing all the virtual machines and data on the SAN.
I finished writing the security policy, which Ed made a prerequisite of me getting to start working on the BladeCenter, so next week I get to start adding operating systems to the blades.
Yay, after almost a whole day of work, I got all the servers involved in DirXML identity management running with new certificates. I also enabled bi-directional synchronization between each eDirectory tree and the enterprise tree, but leaving each local eDirectory authoritative for its “home users”. We have 11 offices, and each office manages its own user accounts, and each office has a unique container (OUs) in eDirectory. We then synchronize everything to our “enterprise directory” which causes GroupWise accounts to get created for every user in our big centralized GroupWise system.
We’ve been starting to need user authentication to eDirectory trees besides our home eDirectory and the mail system, so now that synchronization is complete, all user credentials are replicated in all offices, and with a bit more tweaking, all staff will be able to use the same login ID in any office they go to.
I spent the day futzing with DirXML driver settings and certificates in eDirectory.
We have two offices in Saskatchewan and they do a lot of project-collaboration. Lately, they have a sizable job with the staff split half between Saskatoon and Regina. Instead of forcing them to maintain two project document repositories, one in each office, and then trying to merge them at the end of the project, we are trying to give them access to the master repository in Saskatoon, even for workers in Regina.
To make that work the users from the eDirectory in Regina need to be copied to the eDirectory in Saskatoon, so that they can login in Saskatoon and get access to the directories. I have already started synchronizing all users across the company to corporate office, so that we can use a single authentication ID for corporate web-based services, so this was just an extension of what we are already doing.
I started configuring DirXML to replicate users from our enterprise directory to the Saskatoon office (and Regina office for good measure) and discovered that one of the SSL / TLS certificates involved in securing the DirXML synchronization traffic was expired, because I had initially created it with too short of a life span. Crud. I had to re-issue certificates to all the servers running DirXML, because once the drivers stop and are restarted, they refuse to communicate when a certificate is expired. Even though the certificate expired months ago, I just noticed, because most of our NetWare servers never go down, so the DirXML systems were still working.
My punishment for setting the certificates up with too short of a lifespan was to have to connect to all our eDirectories and use ConsoleOne on a super-slow computer over saturated DSL links to reissue certificates for several hours. Bleah. I still have more work to do tomorrow on this.
Today we got the last storage shelf for the SAN. That leaves a small handful of drives, some RAM, and our 220 power circuit outstanding, and then we can set up the Bladecenter.
James came in during the evening and rearranged the rack to hold all the gear without being a tipping hazard. Thanks for considering our safety! Power is supposed to be in today. I’m keeping my fingers crossed.