04-15-2014 08:37 AM
I've bumped a Windows IS from post-7.01 to post-7.03 (7.03 bundle plus the latest patches that hit selfsolve last week). Now trying to upgrade any remote Windows client fails with Operation failed with error 1638 claiming another version of the same product would already be installed. AFAIK this is a Windows Installer error code. There is one Windows host where remote upgrade succeeds: The IS itself, which already has the patches anyway.
Anyone ever seen this? Is it a bug in the latest CORE patch? Or is something else borked here?
04-15-2014 10:59 AM
I found a lab case for the 7.01 patch bundle that describes the problem this way:
- remove_patch script should remove all the files which have been copied to the clients with the patch bundle installation.
- Installation of a the patch bundle should work again after it had been removed with the script
What is not clear to me is if this script is part of the patch bundle installation package
fails because previous 7.01 build was different version that the one beeing pushed and since internal build number of 7.01 is actually the same, windows installer detects it as unknown upgrade version.
04-15-2014 12:10 PM
thanks for investigating on this issue.
Bob_Clark wrote:When did you download the 7.03 patch bundle? They may have fixed teh Build version problem since the download
That would have been yesterday ;)
I'm also quite averse to using Add/Remove programs in this context. It would probably work to completely uninstall and then install the clients again, but that is a dangerous PITA process: You have to write down/extract what components were installed and manually select them correctly again, and the uninstall process will (without any way to prevent this from happening lest you shutdown your cell server and wait for nasty timeouts) remove the client from the cell. If you (or some other admin) happens to access a backup spec in this state and Ok a seemingly harmless dialog, you may even lose the entire configuration for this client in that spec (trees, excludes, individual object settings). And while all that may work out for the occasional borked client every other year, it's impossible to do for two-score numbers of clients. Remote update simply has to work. It's also no help that the client installation, when started manually through (c)setup.exe, refuses to do an upgrade. What magic is remote upgrade doing to achieve this, which can't be provided in the interactive installer?
So I still hope to find a way to have this work again. I'll start by establishing a second Windows IS on another client, freshly installed and patched to the state the borked one was in until yesterday. If that can still roll out updates, I'll bump it to 7.03, try again, and then latest patches post-7.03, and try again. I'll report on my findings.
04-16-2014 11:59 AM
and here are the results:
- Installed new Windows IS of vanilla DP 7.00, added the 7.01 bundle on top, added the post-7.01 patches on top of that so as to replicate the state my original IS was in 'til monday. This IS had no problems rolling out upgrades to clients on the same version, as expected.
- Installed the 7.03 bundle on top of that IS. It still had no problems to upgrade clients which are on post-7.01, so I bumped some of them to 7.03.
- Installed post-7.03 patches on top of that IS. Upgrading those clients that were bumped to 7.03 in step 2 would work using this IS. However, upgrading post-7.01 clients using this IS fails exactly in the described way.
So my failure was the optimism to expect slipstreaming super-fresh patches on top of the 7.03 bundle to work (like it did a hundred times before with anything between 6.00 and 7.0x), earlier than anybody else would run into this problem and warn the world. So that's now been my duty. If your clients are on 7.01 with additional patches and you want to bump them to the latest available patch state, do NOT apply DPWIN_00669.EXE (and maybe 00684, I've just verified the first one to trigger the issue) to your IS before you have upped all the clients to vanilla 7.03 first. I also don't know what exact combination of pre-7.03 patch state on the clients is really triggering this, it may be quite rare and I just hid it through an odd coincidence. Needs more data from others. For reference, my excat prior state was:
05-23-2014 06:33 AM
I'm having the same issue. I applied 7.03, DPWIN_00669 DPWIN_00684 to a 7.01 cell manager and get the MSI 1638 on most of the clients.
I applied 7.03, but NOT 669 and 684 to another cell manager and I"m able to update the client without errors. It seems like HP screwed up the package code and/or build versions in either DPWIN_00669 or DPWIN_00684.
06-27-2014 07:37 AM
I would say to everyone who contributed to this thread to get a case open to report this problem. I suspect that this will need to go to the lab, and I can't do this from a Forum thread