I am certainly happy about any "evolution" of your software. But I have the impression that you did not understand my complaint. Since we can roll back our environment to any date within the last three months, we would have expected the suggestion/comment: "OK, there is a problem. Please roll back - we have a new installer that takes into account more scenarios of failure like ..

"When you run setup.exe, if first installs the MSI and then goes through these steps:

PreDBConfig

DBConfig

PostDBConfig

PostAddOn

PostMSI"

"

Did not happen.

Considering the "perfect installer". Well every installer is perfect - as long as he just installs. The only thing that matters (I agree with you) is taking eventualities of failure into account. But I think that on a pure LANDesk system (started with 8.6) this should be your job. If this is not taken into account, than this "illness" maybe should be called "experimentitis to customer's" . We know this really well from Microsoft - they also call it  "software evolution"

Damir Bozic, Dr. rer. nat, eidg. dipl.Inf.

Department of Materials, ETH Zürich

http://www.ethz.ch

It is a very logical argument, and one based on reality.

Let's take Microsoft and Linux OS's as a comparison (not to compare against one another - just to give you an idea of things).

- Linux is open source

- Microsoft's OS's are not.

Both have major releases every X months/years (same as LANDesk).

Both have patches coming out for them after release (same as LANDesk).

Why do I compare Linux and Microsoft?

- Linux distro's have a limite amount of funding available for testing and validation.

- Microsoft has no doubt one of the biggest testing and validation departments in the whole world.

Does having a massive testing & validation budget reduce the need for hotpatches to be developped by Microsoft? No - it does not.

Because a testing/validation cycle can only cover so much. Any software vendor tries to nail as many things as possible in testing in validation - if they did not, they would not be able to survive, and half the product would likely not work. Anything beyond a "Hello World!" program is going to have bugs in it.

If you have several million lines of code, it's inevitable that you have bugs. Then there's also things that aren't really bugs (in terms of programming errors) but problems that come to light only if you have environmental things X, Y and Z combined together (say for instance IIS 5, running Oracle 10g and "MyBlueColourScreen" installed on one Core).

EVERY software out there need to have a patching cycle as part of its lifestory. If you believe that the presence of hot-patches is the sign of insufficient testing, then you have either no true experience in the software industry, or are an idealist but do not know it. It would be NICE if software did not need patches - we all would like that. However, that's not going to happen. Any product - mechanical or software - needs maintenance. Because no matter how much you test, it's simply impossible to have ALL the possible combinations covered that are out there.

hartwurz - if you have a personal grudge about an issue hanging around (which you seem to) - please feel free to escalate to the respective regional manager. There's processes in place for dealing with these things, and doing it on the forum is not the best way, as it makes you look as if you have a specific agenda, and it undermines any willingness from individuals to help you - why should I spend time on this?

So - yes, you have very idealistic views on how software should be. That's nice - but that's idealism - that's not reality, as the reality of any software product is that it never is prefect, nor will it ever be perfect. We're all human - we're all flawed - and since there's not a lot of money to be made with a "Hello World!" program (and heck, you can have bugs in that), this is the way to be.

The ONLY alternative would be to NOT release patches. And that is NOT an indication of a 'healthy' product. That is an indication of a slacking sustaining group.

There always WILL be problems - that in itself is not the issue. The issue that a software vendor (or physical product vendor) needs to address is how they're going to deal with those. Patches being released (whether by Microsoft, your Linux distro, or whatever software vendor) are a sign that they're trying to take care of their customers and their product.

It's when you have only silence in that regard that you have REAL reason to worry.

Last word on this from me in this thread.

Paul Hoffmann

LANDesk EMEA Technical Lead.

Huups, sorry but I did not want to grudge around. But I was thinking that this is a public forum, and as I was currently evaluating 8.8 and have something from your regional sales manager I was thinking if  I realy shall invest in a Support Contract. That was my querstion. 

Well  anyway, I am just imagining what my CIO would say if I tell a customer: Hey dont hang around  and grudge.  If you have something to say contact your regional manager.

Maybe somethibg like: Do moechd oims Fiedele schwaetze. 

This is Schwaebisch and if you need a translation contact MY regional sales representative. And all the best with your CIO

Yes - they will need to be upgraded. Makes sense - can't just upgrade the Core, client-components also need updating right :).

We can't just "magically" make the clients do new tricks :).

And you do it just in the normal way - a client deployment. Since you already have LANDesk out, this can take a multitude of forms (self-contained agent) ... there's no "cut and dry" answer for this, as it depends on your environment. I'd suggest looking at the notes on how you do agent-deployments at the moment / discuss it with your ESP, who is familiar with your environment :).

Paul Hoffmann

LANDesk EMEA Technical Lead.

Seems like you have the agent installed on the core. Review this link.

That depends on what you use your Core for, really.

The way to find out, is asking yourself - "what would be my problems if I cut the network cable to the Core" ?

Depending on what you use the Core for (and how heavily), the prime candidates that come to mind are (in no particular order):

- You're no longer able to remote control (well - you could work around that, but if you're using Certificate based RC, the Core MUST be up).

- You're no longer able to do software distribution (push + policy).

- Your clients may not be able to talk to the Core for AV stuff (they can still update their definitions on the WWW as a failover though).

- You're not able to do patch management.

Generally, I'd discourage trying to do this during working hours. Not for a scientific reason,. but "real world experience" leads towards it just being "more likely" that the gremlins of IT conspire against you during working hours. That said, i've done a fair few projects during normal working hours/days as well, and that worked too.

The most important thing is preparation really.

1 - Have a complete backup (of the DB and the Core). So in a "worst case scenario", you can roll back to a state as if nothing had happened.

2 - Test beforehand. If you can (and have VMWare), create a VMware image of the Core (real easy now with the new tool), and test the upgrade in a "safe" environment first.

If you run into problems, take note of what needs to be done, look at logs, etc.

The more preperation is done (or - at the very least - if you have a backup (better still - two) to go back to), things usually go pretty smooth.

It's when people forget their backups (or - what's worse - only make the one and don't check it for being good/healthy - it's not fun finding out your backup's no good when you need to rely on it ) or generally skimp on the preperation that things go wrong.

That's not something that's specifically LANDesk - I've observed this to be true for pretty much any server-related operation. The old adage of "If something can go wrong it will" holds considerably more true when you've got no safety nets :).

Hope that answers your question satisfactorily.

Paul Hoffmann

LANDesk EMEA Technical Lead.

You should have the power to delete your own posts.

If not, let me know and I can delete them for you. Just before I do - that would be the three posts above mine?

I.e.:

The post containing:

""

Sat, 26 Jul 2008 12:35:46 Main install failed. MSI return code is 1603

Sat, 26 Jul 2008 12:36:50 Returncode: -1918091262

""

The post containing:

""

Failed to backup ntstacfg.ini in D:\Program Files\LANDesk\ManagementSuite\ldlogon : (5)

Action ended 13:42:49: LDBackupStaCfgFiles. Return value 3.

MSI (c) (4C:68) 13:42:49:837: Doing action: SetupCompleteError

""

And the post saying to ignore the above two?

Paul Hoffmann

LANDesk EMEA Technical Lead.

Done.

I've also contacted Beau to check into why you aren't able to delete your own comments.

Paul Hoffmann

LANDesk EMEA Technical Lead.