Please help!

Scenario: SCCM 2012 R2 CU5, User Driven Task Sequence, all that is entered in the wizard is computer name and a few variables are for conditional groups. Multiple "Install Application" steps some of which have the maximum of 9 in the list and some which only have 1. Some of the ones with 1 in the list do a restart before and after because these are fairly large applications. An example being AutoCAD 2015. I know the applications install as they have done previously from a task sequence. Nothing has changed with the applications. When running on two physical machines I am getting exit code 16389 on a random and different "Install Application" set on each physical machine. When looking at "History of a task sequence on a computer" I can see the following entry for matlab.

The task sequence execution engine failed to install application that was specified in the 'Install Aplication' action.	16389	Matlab R2015aScopeId_405997B8-D50C-429C-84D6-012DDAF9BDB5/Application_67d17ef4-6423-45c8-bdf8-4e52bd648cc4		

A few rows down in this report I can then see this (see bold and underlined)

... 00
Execution status received: 4 (Application failed to install )
Setting TSEnv variable '_TSAppInstallStatus'='Error'
Setting TSEnv variable 'SMSTSInstallApplicationJobID__ScopeId_405997B8-D50C-429C-84D6-012DDAF9BDB5/Application_67d17ef4-6423-45c8-bdf8-4e52bd648cc4'=''
Completed installation job.
Step 2 out of 2 complete
Sending error status message
   Setting URL = https://server.address, Ports = 80,443, CRL = false
   Setting Server Certificates.
   Setting Authenticator.Set authenticator in transport
Sending StatusMessage
Default CSP is Microsoft Enhanced RSA and AES Cryptographic Provider
Default CSP Type is 24
App install failed.
Install application action failed: 'Matlab R2015a'. Error Code 0x80004005
Setting message signatures.
Setting the authenticator.
CLibSMSMessageWinHttpTransport::Send: URL: server.address:443  CCM_POST /ccm_system_AltAuth/request
In SSL, but with no client cert
In SSL, but with no media cert
Request was successful.

In the same task sequence but on another machine matlab installs fine however other applications also appear to fail with 16389 according to the same report. Here is an example.

The task sequence execution engine failed to install application that was specified in the 'Install Aplication' action.	16389	AutoCAD 2015ScopeId_405997B8-D50C-429C-84D6-012DDAF9BDB5/Application_7742f187-3955-41c8-a7dd-a504c61865c7		

and the more detailed section looks like this (bold and underlined again).

... 0000
Execution status received: 24 (Application download failed )
Setting TSEnv variable '_TSAppInstallStatus'='Error'
Setting TSEnv variable 'SMSTSInstallApplicationJobID__ScopeId_405997B8-D50C-429C-84D6-012DDAF9BDB5/Application_7742f187-3955-41c8-a7dd-a504c61865c7'=''
Completed installation job.
Step 2 out of 2 complete
Sending error status message
   Setting URL = https://server.address, Ports = 80,443, CRL = false
   Setting Server Certificates.
   Setting Authenticator.Set authenticator in transport
Sending StatusMessage
Default CSP is Microsoft Enhanced RSA and AES Cryptographic Provider
Default CSP Type is 24
App install failed.
Install application action failed: 'AutoCAD 2015'. Error Code 0x80004005
Setting message signatures.
Setting the authenticator.
CLibSMSMessageWinHttpTransport::Send: URL: server.address:443  CCM_POST /ccm_system_AltAuth/request
In SSL, but with no client cert
In SSL, but with no media cert
Request was successful.

The AutoCAD example above says "Execution Status 24 (Application Download Failed). Whereas matlab was Execution Status 4. I must say that Status 24 is more common error here however we have seen Status 16 as well. It is totally random and on another machine matlab and AutoCAD install fine and another application can fail. Just for reference both Matlab and Autocad are single applications in the install application step plonked in between two restart computer steps.

Does anyone know what might be causing Status 24? I know it says Download failed but a list of what these Execution Status messages mean and possible fixes/troubleshooting steps would be really helpful.

Someone else seems to be having a similar issue to this here: http://www.windows-noob.com/forums/index.php?/topic/12052-random-issue-plaguing-osd-deployments-exit-code-16389/ 

I am not the only one here experiencing this either my colleague in the deployment team has also seen this issue. For reference a lot of these applications install just fine when advertised in software centre.

OSD has become very fickle and it was already quite fickle. It is causing us lots of problems. Stuff that did work suddenly doesn't now. Can anyone offer any pointers on what to check regarding this download failed message?

Double check LocationServices.log. The client might not be able to find and/or talk to an MP right after the restart. Also search for SMSTSDownloadRetryCount, SMSTSDownloadRetryDelay and SMSTSMPListRequestTimeout (task sequence variables) in this context. 

Thanks Torsten I will check the log file. Regarding the variables I have those at the very start of my task sequences like this.

SMSTSDownloadRetryCount = 5

SMSTSMPListRequestTimeout = 15

SMSTSMPListRequestTimeout = 300

This was from a while back and has really been copied through into our builds from something we read on these forums to help with OSD failures. No idea if it is still required in R2 CU5 but they are still there. 

Looking Through LocationServices.log nothing appears to be standing out there are a few red and yellow lines with cmtrace but i think they are normal stuff like "Sending location services HTTP faolure messages" followed by "Successfully sent location services HTTP failure messages" and then "failed to retrieve MP certificate authentication over http". This group appears 3 or four times towards the start of the log. Bit further down there is "1 assigned MP errors in the last 10 minutes, threshold of 5" this goes all the way to 4 and the log carries on. I can see it retrieve the MP name from AD and from the registry.

The Status Message 4 stuff we have figured out, those apps were zipped up with WinRAR and it was reporting an unmatched exit code 1000 (WinRAR adds 1000 to the return code). 

Just the Status 24 messages to figure out. I have pulled a machine from the other side of campus (using a different DP) and brought it to my office to see if it could be DP related.

It it makes any difference we only have 1 MP. We are all scratching heads a bit here. Will post SMSTS.log and any other info from the TS later as its running now.

After the reboot before your application install step, does LocationServices.log indicate that a valid MP was found? There is a known issue that was fixed recently for OSD clients not finding a MP fast enough after reboot.

Also, are you installing CU5 in the client install command line during your task sequence(s)? The issue I'm referring to was fixed in CU4 (more info on the hotfix rolled into CU4 here).  Prior to that, we ran a small vbscript after each reboot that would check for a valid MP, if it doesn't exist, trigger a MP list refresh, wait 15 seconds and try again.

I posted the script previously here.

• Edited by Derek Gary 19 hours 6 minutes ago added hotfix

Hi derek yes we are using the PATCH property to install CU5 when the client installs. I am wondering if adding a ping command line or something after the restart might help?