Error:Unable to reload \\Documents\opbm\running\manifest.xml...

[kathy-smith]

... cannot restart.

Started an official run through the cli. Got the above error message from restarter.exe [Fatal Error] mainfest.xml:2151:267: An invalid XML character ...

[ghost]

Copy manifest.xml to Z: drive.

[kathy-smith]

Debug info copied to Z.

[ghost]

manifest.xml will need to be copied. If it has an invalid character it won't
be loadable into the debug info xml.

Cossatot Analytics Laboratories

-----Original message----- From: kathy-smith
reply@reply.github.com To: "Rick C. Hodgin" rick@canalabs.com Sent: Sat, Oct 29, 2011 02:43:21 GMT+00:00 Subject: Re: [OPBM] Error:Unable to reload
\Documents\opbm\running\manifest.xml... (#104)

Debug info copied to Z.

Reply to this email directly or view it on GitHub: https://github.com/van-smith/OPBM/issues/104#issuecomment-2563187

[van-smith]

The error duplicated on an Official Run made form the GUI.

[van-smith]

manifest.xml copied to Z.

[ghost]

Please also copy out preboot.xml and postboot.xml from the same directory as manifest.xml.

[ghost]

Never mind. I discovered the cause. The preboot.xml filename, in the presence of a space in the user name, was inserting two double-quotes around the name. Fixed in most recent push.

[kathy-smith]

A similar failure has occurred with the latest code pull while attempting to iterate over an Official Run:

_ Loaded edits.xml Loaded scripts.xml OPBM System Initialization completed successfully [Fatal Error] manifest.xml:2151:267: An invalid XML character (Unicode: 0x{2}) w as found in the value of attribute "{1}" and element is "1e".

Error: Unable to reload C:\Users\kathy smith\Documents\opbm\running\manifest.xml manifest file, cannot restart. _

Will delete all local files and try again.

[kathy-smith]

The failure occured a second time. This attempt was run after deleting all of the local files.

[kathy-smith]

The failure occured a third time. This Offical Run attempt was made through the GUI.

[ghost]

It is not running the preboot.exe app for some reason. Please post your debug info data, along with manifest.xml as two separate files to help me to identify the cause.

[kathy-smith]

I was running some other tests. I'll rerun it on the llano right now and
send you the information.

Thanks, Kathy On Sun, 30 Oct 2011 10:30:37 -0500, Rick C. Hodgin
reply@reply.github.com
wrote:

It is not running the preboot.exe app for some reason. Please post your
debug info data, along with manifest.xml as two separate files to help
me to identify the cause.

Using Opera's revolutionary email client: http://www.opera.com/mail/

[kathy-smith]

Just tried an official run. I have put a folder called Issue 104 on the z drive in the defug folder with the manifest xml and debug information. Thanks.

[ghost]

I do not see any content in the "Z:\Amazon S3 US Bucket\debug\Issue 104 Manifest xml" folder, nor the issue 74 or issue 106 folders.

[kathy-smith]

Sorry, let me try to copy them out there again.

Kathy On Sun, 30 Oct 2011 16:09:10 -0500, Rick C. Hodgin
reply@reply.github.com
wrote:

I do not see any content in the "Z:\Amazon S3 US Bucket\debug\Issue 104
Manifest xml" folder, nor the issue 74 or issue 106 folders.

Using Opera's revolutionary email client: http://www.opera.com/mail/

[ghost]

The log reports the adobe acrobat reader uninstall script completed successfully. The script that was running was the firefox uninstall script, which did not complete. Line 8060 shows the script is the one running, and that it is not complete. This would have been the last status update the harness wrote before running that atom, as it does not write content to disk while the script is running, but only after.

Is there a screenshot of the failure visible? It appears the firefox uninstaller process hung.

[ghost]

I think the harness should have a "run in debug mode" which writes items to disk when they are captured (for this purpose and kind of testing), as well as a while-running-written-to-disk snapshot of running processes periodically.

[van-smith]

No. I didn't take a screen shot this morning. Sorry.

You have everything that there currently is about the issue.

On Sun, 30 Oct 2011 16:21:19 -0500, Rick C. Hodgin
reply@reply.github.com
wrote:

The log reports the adobe acrobat reader uninstall script completed
successfully. The script that was running was the firefox uninstall
script, which did not complete. Line 8060 shows the script is the one
running, and that it is not complete. This would have been the last
status update the harness wrote before running that atom, as it does not
write content to disk while the script is running, but only after.

Is there a screenshot of the failure visible? It appears the firefox
uninstaller process hung.

Using Opera's revolutionary email client: http://www.opera.com/mail/

[kathy-smith]

No. I didn't take a screen shot this morning. Sorry.

You have everything that there currently is about the issue. On Sun, 30 Oct 2011 16:21:20 -0500, Rick C. Hodgin
reply@reply.github.com
wrote:

The log reports the adobe acrobat reader uninstall script completed
successfully. The script that was running was the firefox uninstall
script, which did not complete. Line 8060 shows the script is the one
running, and that it is not complete. This would have been the last
status update the harness wrote before running that atom, as it does not
write content to disk while the script is running, but only after.

Is there a screenshot of the failure visible? It appears the firefox
uninstaller process hung.

Using Opera's revolutionary email client: http://www.opera.com/mail/

[kathy-smith]

These are all copied out there now.

On Sun, 30 Oct 2011 16:09:10 -0500, Rick C. Hodgin
reply@reply.github.com
wrote:

I do not see any content in the "Z:\Amazon S3 US Bucket\debug\Issue 104
Manifest xml" folder, nor the issue 74 or issue 106 folders.

Using Opera's revolutionary email client: http://www.opera.com/mail/

[kathy-smith]

I ran this again with the updated code from this afternoon. The first run of the benchmark completed successfully, it rebooted and then failed. The name of the file for results was to be kathy 10305. I will put the screen capture of the restarter.exe, manifest xml and debug file on the z drive.

[kathy-smith]

The folder is Issue 104 - llano official run from GUI

[ghost]

It is the same issue.

Can you verify there are both preboot.xml and postboot.xml files in the c:\users\user name\documents\opbm\running\ directory? And please post their contents to the folder on Z: drive.

[kathy-smith]

I put a folder on the z drive "llano failed run with filename kathy 1013" There were no preboot and postboot files in the running directory.

[ghost]

The preboot and postboot executables are not running properly then, being the recurring root cause of the manifest.xml reload failure. While the failure is recorded immediately after the first reboot, manifest.xml remains in memory for the remainder of the pass on an Official Run. Once it reboots the second time, it tries to load the disk-based manifest.xml back into memory, and there encounters the failure in the load parser.

I'll have to create some custom tests to run on that machine to figure out why it's not working.

[kathy-smith]

Okay, the file name is "llano failed run with filename kathy 1031" Sorry for the typo

[ghost]

Can you verify that preboot.exe and postboot.exe exist in the ...java\autoIt\common\opbm\exe\ directory? And that they are 109KB files? And if so, please click on the path area at the top and paste its value here. For the dev machine, it's "C:\cana\java\autoIt\common\opbm\exe", for example.

[kathy-smith]

Okay, I found the preboot and postboot exe and they are each 109KB. I don't understand what you are asking me to do next? Click on what path area? On my pc? my path is c:\my opbm\java\autoit\common\opbm\exe

[ghost]

Yes, that's the path I needed. I'll try launching from that directory to see if I can replicate the failure.

If you're feeling able to do this, here are some steps to try to see if it's running properly. Read all steps before you do any of these (and make sure you are running with the java.exe command line window visible, and not just executing the opbm.jar directly, you'll need it to terminate the app early). Do each of these steps quickly:

1) Go to Atoms -> and click "Standalone Reboot Time" and click Run Atom 2) It will begin to run and say it's rebooting. 3) Quickly, go to the Windows Start button and type "cmd" and press enter. 4) In the command window, type this: shutdown /a

• Note: A toast message should appear which says "Shutdown has been canceled" or words to that effect. 5) Quickly, before OPBM times out, click the "X" in the java command line window, this will terminate OPBM prematurely before it times out and wipes registry key entries.

Once this is done, the system is in the state it will have been in when Windows began the shutdown. Here's what you're looking for:

Go to Task Manager, Processes, sort by process name, look for "preboot.exe" running. Note if it is. Go to the opbm\running\ directory and note if there is a preboot.xml file, and if it is being updated once a second. Go to Windows Start button, type "regedit" and locate these registry keys:

HKCU\Software\Microsoft\Windows\CurrentVersion\RunOnce\opbm HKCU\Software\Microsoft\Windows\CurrentVersion\RunOnce\opbmpostboot

Copy and paste their values here, by right-clicking on "opbm" and "opbmpostboot" and choosing "modify" and copying the contents in "Value".

This will help me find out if things are running properly, or why they aren't running.

[ghost]

After you have done these steps, manually terminate the preboot.exe process, and manually delete the opbm and opbmpostboot registry keys, or launch and immediately close OPBM once (to wipe them through OPBM's normal shutdown facilities).

[kathy-smith]

Sorry to respond so late. I will try this tomorrow

I have done a lot of iterations of install/uninstall combinations today with no failures.

I can see when there is a failure and the postboot and pre boot exe's are not being generated.

I'll try to capture the info on the systems tomorrow.

Kathy

On Oct 28, 2011, at 9:50 PM, Rick C. Hodgin wrote:

manifest.xml will need to be copied. If it has an invalid character it won't
be loadable into the debug info xml.

Cossatot Analytics Laboratories

-----Original message----- From: kathy-smith
reply@reply.github.com To: "Rick C. Hodgin" rick@canalabs.com Sent: Sat, Oct 29, 2011 02:43:21 GMT+00:00 Subject: Re: [OPBM] Error:Unable to reload
\Documents\opbm\running\manifest.xml... (#104)

Debug info copied to Z.

Reply to this email directly or view it on GitHub: https://github.com/van-smith/OPBM/issues/104#issuecomment-2563187

Reply to this email directly or view it on GitHub: https://github.com/van-smith/OPBM/issues/104#issuecomment-2563212

[ghost]

Testing on local SUTs.

[van-smith]

Verified. Specially configured Llano SUT (space in user name, space in OPBM file path) passed two Official Runs.