Saturday, November 04, 2006

 

Updates coming for server issues.

Not much updates lately due to a chain of events that brought down my test server and not able to test a few updated files from the Livestate mail groups. But from the Livestate support, there is an update coming soon for the server no response issues in either DOS or package pushes. As a matter of a fact, a new ccmmsg.jar is going to be tested on our server to check for improved server stability after big package jobs.

Coming up I have a class scheduled for the Package manager. The renewal of the contract came with 2 classes, one is coming up next week which I have pretty much been told is too late to go. Which is a shame since I have been asking about this for 2 month now, and now I have to feverlishly finish paper work to get it approved.

Thats just the crappy part of working with a .edu, so much politics but not very much progress can be made due to those.



Tuesday, October 17, 2006

 

Another step.

Walked into an interview at my work place for the Sr Systems Engineering position. Walked out surprised to hell. Not only is there good news for me, but 6 other people will get promoted or hired for the new positions created.

Not that I feel like I deserved it, because I don't. Livestate being the main reason of the promotion, still has alot of bugs. Although it is a working product vs when I first got on it I'm not sure I see it as a reason for the promotion. Maybe it's just the more mature way of looking at automated solutions rather than just man power.

Got a few files from a fellow Livestate admin that I will have to try on my servers to findout if they will help the stability of them. The issue with machines not always connecting on the first try will get looked to tomorrow morning, and more Livestate E sessions to find out the issues with Co-op image capture and replication.

Thats all for now, hopefully the updated db1.jar and the Ovl file will provide more stability for the Tomcat services.

Saturday, October 14, 2006

 

Server and client communication

Fri the 13th went exactly like how it would if the ghost of the past had anything to do with it. I revisited a campus that had some machines imaged via Livestate and the status showed either in progress or waiting. And the job was scheduled about 3 weeks back all the users had to do was turned the machine on to initiate the process.

It seems like the first or the several subsequent reboots does not connect the machines to the Livestate server. It took sometimes 6-7 reboots for the machines to even see it's ID on the server and start the image jobs. A few machines did not even connect at all even with the jobs scheduled, and with only about 6-7 concurrent sessions.

So this brings up an issue of whether it's the Server side Services thats at fault like I previously thought. Or is this more of a bug within Livestate software it's self.

The good news seems to be one of the other companies had a visit with some Livestate Engineers and seems to have alot of their server client issues resolved through a new OVL, db1.jar and yet another new agent. So maybe Monday would be the day to try these new files if he's willing to share.

: )

Also I printed the scheduled training sessions in Santa Monica for Nov on my boss's desk today with a sticky inquiring the status(Whether I can go, or did they have other plans). Remember this is for Nov's first week and last which is 3 weeks away! Working for this company thoroughly frustrates me as the upper management have no visions on how things should be done and rather deal with one emergency at a time due to a lack of update and maintenance on current projects and employee development.

Thursday, October 05, 2006

 

Livestate, Past and Future.

Just when I thought when I had turned this project around atleast to the point of some predictability with the workarounds. Today, more odd Livestate Delivery behaviors, which will undoubtedly result in the weekend spend fruitless researching.

Last night we had a desktop my boss asked me to re-image with Livestate at a remote location. So I terminal in to the machine, and run the SSM HP bios util to change the boot order, then install the agent via script. After that, everything heads in to the twighlight zone.

First, the machine just reboots and goto some error about the U: SW which has to do with error mapping the drive, and reboots. Normally, I restart the service on the server, and sometimes remove the client and add in manually. Except this time When I click on the machine in the CC, everything freezes, and I continued to get that U:/ sw error. So I try to delete the OS jobs assigned or delete the machine from the CC I get (Unspecified Error). I had to manually stop the service on the server and remove the folders for the machine to delete it from the server. So I added the machine back after a server reboot, and still the same error. I add another machine to the server and image starts to deploy fine, then all the sudden it goes to a cant find files error 15 minutes into the job. But a reset on that second machine allowed it to finish the job. First machine still gets the same errors at this time.

Also at this time I am building a Livestate based image. Which for us involves a Livestate OS install, and then build upon it and then use the Image Prepare Image to capture it for redeployment. So I get everything configured and starts the image capture process(During this it seems to not be able to reset the local admin account to blank) half way through the image, it errors out(Not enough space left on destination drives). This is on a HP DL380 G4 with about 418 gigs of free space on the array. Bascily that image is trashed and I had to start all over, in the mean time, the next 2 image capture for other images went along just fine.


I'm not sure whats going on, but now I am less frustrated as I found other folks with the same issues. And these guys are much better prepared than I am. They not had Symantec Engineers on-site a few times to help with the issue, they are the reason for the 6.011 patch, and not much works right for them still. All the issues I am having, they have more. And they only have 5,000 machines!!!! So the product develop cycle isn't very flattering at this point. The only thing that stings me is that the Support told them that they were an "isolated issue".

So barring a miracle that will come with Version 7.0, now developing in India. I am going to have to ask for a raise, this is getting silly trying to make a product this unreliable work with such a huge network when people with simpler arrangement has more issues. This is beginning to not be realistic in relying on this product to work in an enterprise environment.

Tuesday, October 03, 2006

 

Woeful tuesdays....

Might have to call it Black Tuesday. Rumor is going around that our big wig is forced to retire and more changes are coming our way, I guess this means no promotion this X-mas. Ah well, life goes on.

I never did get a summery of whats done in terms of planning and development in the anticipation of rolling out Livestate Delivery to the real environment. So here is the important things we have done.

1. Divided up all the sites in to 5 blocks of roughly equal sized number of total machines. The choices were based on machine types to keep each managed sections as homogeneous as possible for drivers and hardware updates in the future.

2. Build a test server to test images and software prior to roll-outs.

3. Livestate images built for 90% of inservice PC types.

4. Web interface to allow heldesk to reset/reimage machines remotely.

Today I tested more setting changes with some help from Symantec. Mainly in the form of some changes in the ccmad configs for the agent behavior(Port 8443 and https as defualts) to try to eliminate the machines joining one or more server upon add by script or CC. There is also an issue of machines that will not allow Livestate to install the agent when deployed via the CC( cannot connect to remote machine, possible credentials). This applies to both Auto Add and Auto Discover.

But with the CCMservers set to only having the setting of the machine it resides on(SLDserver="10.192.168.1") all seems to work wonderfully. This is with a CCMservers.cfg injected into the local _integra folder. Which theoretically should allow the machine to be moved to other servers tats on that list and truly poll from the new server it now resides on.

So will Nov bring the training thats been promised all this time? I hate to think all this work is going to be a waste once the new management comes in.

Monday, October 02, 2006

 

Technorati Profile


 

Livestate Delivery Product 8 month summery

This blog will be used to document my experience with the Symantec Livestate delivery product. If your unfamiliar with it, www.symantec.com has some info on the LSD product line and it's uses in desktop management.

Most of what we have done with the product up to this point has been documented in my blogs at:
http://blog.myspace.com/nukeitall

To sum it all up, we have had the Livestate product for about a year, and I have been using it for about 8 month now. First few month was rocky at best. The product had very little documentation and was buggy at best. This couped with our managements decision to not pay for training led to the Senior Engineer of the project to resign. After alot of work and digging around online at the Icommand website I've only got a few of the functions to work, and thats not with any consistency. And all this was working 10+ hours a day including weekends and some nights.

After countless time of speaking to the management about training and implementation of the product, we got a 2 day on site with Liz Alfond of Symantec. The under lying of the product is so vast there would be no way for anyone to efficiently setup this product and work with all the different configs with out the Symantec professional services. After all, our environment is 17,000+ desktops and 50+ remote sites in a MAN type network setup across lots of Vlans.

After solving alot of the minor issues we still had after all this time, today was the first day I was able to deploy the agent to a full lab using the CC and Auto Discover, and deploy a software package with 100% success! This feels really good after the management basically put this project to doom from the beginning with the lack of planning, training or manpower.

As of now, we have just signed a 3 year renewal with the product which will include 2 weeks of training on the Livestate Delivery and the Package manager parts of the suite.

So after the 8 month, here is what works:

Livestate Image delivery - unmanned computer deployment by just knowing the Mac addy and machine being put in network boot. The machine will load the image and join the machine to the domain without any user interference.

Windows OS install - unmanned OS installs.

Software deployment - Agent deploy by logon script or Command Center deployment.

Image creation - from Consistent mode machines.

Things that is buggy or not working yet:

Patch manager - Will not DL certain patches.

Coop mode image capture -Since we have lots of pre-Livestate images with software configured.

Replication - Very buggy. Have open issues with Livestate on this.

Livestate Service - Seems to have some issues after large #'s image deployments and have to be restarted.


Time will tell how this project will go since the lack of training and understanding of the product by the current staff or the management seems to be the big issue at hand. And the fact that we do not even have a department structured for packages tobe deployed is going to be a huge mistake.


From now on, all testing and results will be posted here and no longer on the old forums.

This page is powered by Blogger. Isn't yours?