When our installer getting failed, a rollback process is starting by default.
Because our installer involves Windows services, we use the Optional Rollback Script property to close and delete these services in case of a failure after their installation.
Our problem is that some processes remain assigned to some ports, a situation which leads the user to not be able to delete the installation directory without killing these ports first.
Now, in the command line, it's quite easy to locate and kill these processes with the commands:
netstat -ano | findstr :<portNumber>
taskkill /PID <processId> /F
My problem is that I'm not able to run these commands through the Run Script action, getting their processes id's and than close them. (not matter what, I cannot get the output back)
Is there any such a build-in option in the Installer? If not, is there any alternative way?
You can use the WinProcesses API to kill processes:
https://resources.ej-technologies.com/install4j/help/api/com/install4j/api/windows/WinProcesses.html
As for calling netstat, you have to call
netstat -ano
with a "Run executable or batch file" action and set its "Redirect stdout" property to "To installer variable". Then you can parse the output in a "Run script" action.
Related
I have a supervision tool that can deploy scripts on customers end devices.
I'm trying to make two powershell scripts.
The first one is supposed to launch a "chkdsk disk_name: /f /r".
The second one is supposed to extract the result of the chkdsk after the reboot from the event viewer.
The second script is operational. My problem is with the first one.
I think that when I'm launching my job from my administration tool, the script is launched on the end device, but when you type "chkdsk disk_name: /f /r" on a command prompt, it asks if you want to do the chkdsk at the start of the machine because the disk is actually in use. I think that the letter "Y" that you have to type to confirm, is blocking the execution of the command (and my script by consequence).
I didn't find in the documentation of the command any method to launch it with a "default confirmation".
Do you have any idea of what I can do to automate this?
Sorry for my English, it's not my native language.
Thank you all!
I tried to launch the script (it's in admin mode when my administration tool launch it's job) but the result was that my job was running indefinitely and at the restart of the machine, the check disk is not performed.
We have an application server running as a service, when some configuration is loaded it starts a bat script which has to run the powershell command Stop-ClusterGroup DRMSERVICES and then start it again.
The bat file works flawless when I manually execute it by dobbelt clicking. But when the service is running the bat, it does not finish, or execute the powershell command.
Bat file looks as follows
#echo off
powershell -command Stop-ClusterGroup DRMSERVICES
powershell -command Start-ClusterGroup DRMSERVICES
The service runs the bat file in silent mode, as a main difference.
I have tried with various switches including the -ExecutionPolicy Unrestricted and START /wait etc
Creating a seperate ps1 file and have the bat execute this instead.
All with the same output:
Manually executing the bat works
When the service executes the bat, it does not work.
I know the bat file is executed by the service, as inserting NET STOP servicename is working correct.
In the powershell event viewer I can also see event of the powershell commands take place.
The difference between manually executing and have the service execute the command in the event viewer, is event id 800 which states info about 'execution pipe' this is not present when the service is executing the bat.
The service does not wait for the powershell, and thus it does not have time to stop the cluster before exiting.
I'm lost whether this is a permission issue, syntax error or whatever.
Hopefully somebody can help
UPDATE:
I have tried with all proposed solutions, all with same result, the bat file works when double clicked, but the service does not execute the powershell command. Pure cmd is executed, as I can pipe to a txt file. I even got to a point when trying runas that the output log text wrote "insert administrator password"
I even managed to have our software guy change our software to call a powershell directly instead of a bat, same result. Powershell won't execute the command, this tells me it probably is permission, but everything have been set to log in as admin and run as admin for the sake of success, but still nothing.
I solved the problem.
Because the service is a 32bit process, it will execute a 32bit powershell.
FailoverClusters module only exists as a 64bit module.
By using %SystemRoot%\sysnative\WindowsPowershell\v1.0\powershell.exe
The service is able to open a 64bit session, and thus use the failover cluster module.
As a side note, the sysnative folder is only visible from a 32bit session, therefore it cannot be found via browsing in a 64bit os.
I think i have dealt with this kind of issue before, after the,
powershell -command Stop-ClusterGroup DRMSERVICES
you need to have cmd wait for a certain number of seconds, and then test if the DRMSERVICES is now stopped, if it is stopped then to start the DRMSERVICES again. This way cmd will keep waiting, and then check if the service has stopped.
After a certain number of tries, maybe have a way to stop checking and exit the script, for example it is trying to stop the service, and has run into a problem.
There is a timeout command in cmd
I have a scheduled task on a Windows 2008 R2 server. The task includes a Start In directory entry. The task runs, and the batch file it runs does what it is supposed to do. When I run the batch file from a command prompt, I see no errors. The problem is that the "Last run result" is 0x1 (incorrect function call).
I did get this at one time with an incorrect DOS statement IF EXISTS file.txt DO (Copy file.txt file1.txt) that was corrected by dropping the DO statement. The current batch file does not show me any errors or warnings.
Why am I getting a 0x1 result?
Batch file that is run:
PUSHD \\JUKEBOX4\Archives\CallRecording
REM only move csv and wma together. wma should be created last.
IF NOT EXIST C:\CallRecording (MKDIR C:\CallRecording)
FOR /f %%f IN ('DIR /b *.wma') DO (
IF EXIST %%~nf.csv (MOVE /Y %%~nf.* C:\CallRecording\)
)
POPD
CD /D "C:\Program Files (x86)\Olim, LLC\Collybus DR Upload"
CollybusUpload.exe
POPD
Info on scheduled task setup:
Program to run: C:\Program Files (x86)\Olim, LLC\Collybus DR Upload\CallRecordingUploadFromH.cmd
Start in: C:\Program Files (x86)\Olim, LLC\Collybus DR Upload
Run whether user is logged on or not, highest privileges.
History screen, task completed entry
"Task Scheduler successfully completed task "\Call recording upload to portal from NH" , instance "{1449ad42-2210-427a-bd69-2c15e35340e6}" , action "C:\Windows\SYSTEM32\cmd.exe" with return code 1."
First screen of Task Scheduler shows "Run Result" of "Success"
It seems many users are having issues with this. Here are some fixes:
Right click on your task > "Properties" > "Actions" > "Edit" |
Put ONLY the file name under 'Program/Script', no quotes and ONLY the directory under 'Start in' as described, again no quotes.
Right click on your task > "Properties" > "General"
| Test with any/all of the following:
"Run with highest privileges" (test both options)
"Run wheter user is logged on or not" (test both options)
Check that "Configure for" is set to your machine's OS version
Make sure the user account running the program has the right permissions
I found that I have ticked "Run whether user is logged on or not" and it returns a silent failure.
When I changed tick "Run only when user is logged on" instead it works for me.
I've had the same problem. It is just a batch-file, working when manually started, but not working as a scheduled task.
there were drive-letters in the batch-file like this:
put z:\folder\file.ext
seems like you should not use drive-letters, they are bound to the user, who created them - for me this little change made it work again:
put \\server\folder\file.ext
For Powershell scripts
I have seen this problem multiple times while scheduling Powershell scripts with parameters on multiple Windows servers.
The solution has always been to use the -File parameter:
Under "Actions" --> "Program / Script" Type: "Powershell"
Under "Add arguments", instead of just typeing "C:/script/test.ps1" use -File "C:/script/test.ps1"
Happy scheduling!
Windows Task scheduler (Windows server 2008r2)
Same error for me (last run result: 0x1)
Tabs
Action: remove quotes/double-quotes in
program/script
and
start in
even if there is spaces in the path name...
General:
Run with highest privileges
and
configure for your OS...
Now it work!
last run result: The operation completed successfully
Probably not the cause of the OP's problem; for me the problem was caused by the fact that my program called a SQL function, and the service account the windows task was set up with did not have the required SQL permissions. That also gives a 0x1
This answer was originally edited into the question by the asker.
The problem was that the batch file WAS throwing a silent error. The final POPD was doing no work and was incorrectly called with no opening PUSHD.
Broken code:
CD /D "C:\Program Files (x86)\Olim, LLC\Collybus DR Upload" CALL CollybusUpload.exe POPD
Correct code:
PUSHD "C:\Program Files (x86)\Olim, LLC\Collybus DR Upload" CALL CollybusUpload.exe POPD
In my case it was an encoding issue. We wanted to start en existing batch file, and it resulted in "return code 1", and the desired action wasn't performed. I've accidentally found that the batch file was shown in Notepad as one with UTF-8 encoding (actually without any reason, as we have no special characters in the text). I saved it as ANSI, and it solved the problem for us. Might be, that it was a kind of encoding corruption in the file that prohibited Task Scheduler and cmd.exe to open the file, although it was displayed correctly in Notepad.
On our servers it was a problem with the system path. After upgrading PHP runtime (using installation directory whose name includes version number) and updating the path in system variable PATH we were getting status 0x1. System restart corrected the issue. Restarting Task Manager service might have done it, too.
I was running a PowerShell script into the task scheduller but i forgot to enable the execution-policy to unrestricted, in an elevated PowerShell console:
Set-ExecutionPolicy Unrestricted
After that, the error disappeared (0x1).
Just had the same problem here. In my case, the bat files had space " "
After getting rid of spaces from filename and change into underscore, bat file worked
sample before it wont start
"x:\Update & pull.bat"
after rename
"x:\Update_and_pull.bat"
For me the problem was the PowerShell script being ran had #Requires -RunAsAdministrator at the top, meaning it needs to run in an elevated command prompt as an Admin, but the user the Scheduled Task was set to run as wasn't an admin on the local computer. So even though Run with highest privileges was checked in the scheduled task, I still had to make the user an Administrator on the computer. Once I did that, the script ran as expected.
Since there is always more than one reason this could happen I thought I'd share some troubleshooting tips that helped me diagnose my issue.
Always adding a "start in" parameter first since thats an easy fix, even just adding the drive letter can help, e.g. C:\
If you're running "whether user is logged on or not" and it is failing it might be an issue with your user and/or user environment.
Switch the task to run only when user is logged in temporarily for
troubleshooting purposes.
Make sure you're actually logged in AS the user you're telling the task
to run as. (PATH and other environment variables are different by user
and if you see the task running on one user successfully that doesn't
necessarily mean it will run successfully for another user even if they're in the same security group.)
Add pauses or some other type of debugging to your script to give you
time to see any errors that may pop up.
Perform a manual run from the task scheduler window.
Fix any errors you see from your debugging statements. Rinse and repeat.
If it runs successfully switch back to run "whether user is logged on
or not" and try another manual run. If it works now you're all set.
If nothing has helped so far you might need to dig in deeper to your user and file privileges. My troubleshooting tips assume that you have been able to get a past task running using a specific user login already. They don't cover building a scheduled task from a fresh install necessarily. Luckily I haven't had to do that.
What solved it for me was that I was using a local administrator account instead of the domain account so I changed the "Run as" to the domain account.
It turns out that a FTP download call using winscp as last thing to do in the batch caused the problem. After inserting the echo command it works fine. Guess the problems source could be the winscp.exe which do not correctly report the end of the current task to the OS.
del "C:\_ftpcrawler\Account Export.csv" /S /Q
"C:\Program Files (x86)\WinSCP\WinSCP.exe" /console /script="C:\_isource\scripte\data.txt"
echo Download ausgeführt am %date%%time% >> C:\_isource\scripte\data.log
I am writing a batch for a new deployment of my company's software.. Here is what I have so far...
wscript.exe "invisible.vbs" "apache_start.bat" /wait
wscript.exe "invisible.vbs" "mysql_start.bat" /wait
"C:\Program Files\Internet Explorer\iexplore.exe" http://localhost
So as you can see, this script should start apache, then start mysql and then open the default page with IE.
The problem is if the user runs this script twice, it runs apache and mysql twice and loads two seperate instances. The solution I need is a way to check to see if the processes are already running and, if not, run the two wscript commands. I am absolutely horrible with shell, so please try to give specific responses! I am a software engineer, not a sysadmin. Thanks for the help!
As a software engineer I think you have a leg up on scripting over some sysadmins...
Using PowerShell would make this easy. Use the following template to execute the services - you'll need to use it twice, and follow up with launching IE as above.
If ((Get-Process mysqlprocessname.exe)) {Write-Host Skipping MySQL}
Else { Start-Process ...}
This is going to take a few minutes for you to research the best way of starting a process with PowerShell. Also, you might want to pipe Start-Process to Out-Null so the script waits to start IE and Apache.
Others may want to chime in with a simpler way from a batch file.
For XAMPP, there is a pv.exe file in the apache/bin folder that XAMPP uses to see if a service is running. Look at WorldDrknss' answer in this thread for some great info: http://www.apachefriends.org/f/viewtopic.php?p=80047
The code to solve your problem is to modify your mysql_start.bat file to this:
#echo off
apache\bin\pv mysqld.exe %1 >nul
if ERRORLEVEL 1 goto Process_NotFound
echo MySQL is running
goto END
:Process_NotFound
echo Process %1 is not running
mysql\bin\mysqld.exe --defaults-file=mysql\bin\my.ini --standalone --console
goto finish
:finish
That will check if mysqld.exe is running. If it is, it just echos that out. If not, it starts the service.
When I'm writing a batch file to run automatically, how do I write it so that when the batch file is run, it can pause for a couple seconds in between commands?
Context:
psexec \\server -u user -p pass cmd
[there needs to be a pause here for psexec to establish a connection]
dir /s >output.txt \\server\shared
*Note: the reason I run the dir command server-side using psexec and not locally is because it's much faster to run dir on a local machine than remotely, and time is of the essence.
When I'm doing this by hand it's obviously easy, I just wait. But running a batch file makes it run all commands at near instant speeds next to each other, regardless of the completion status of the last command. How do I put in a pause?
On Windows Vista / Windows 7 you can use the timeout command:
timeout /T [delay in seconds] /NOBREAK > NUL
On previous versions of Windows, you can use the ping command (the ping command has 1000 ms of delay between each iteration):
ping -n [delay in seconds + 1] 127.0.0.1 > NUL
Some versions of Windows (like Windows Server 2003) has the sleep.exe executable:
sleep [delay in seconds]
Note: Windows Resource kit for 2003 contains sleep.exe command.
If you don't know the Windows version, simply use the ping hack since it'll be available.
There is timeout command in more recent version of Windows:
timeout /T 10
Windows Resource kit for 2003 will install on Windows XP. It contains SLEEP.EXE which can be used from a command batch file.
download is here http://www.microsoft.com/download/en/details.aspx?id=17657
I think the information here: http://malektips.com/xp_dos_0002.html would explain it better than I.
There's still the case of error handling though (what if the remote machine isn't up?). cmd.exe is quite useless for doing any remote activities for the most part, using powershell would enable so much more.
EDIT::
In fact, you can execute a program stored locally with psexec (it gets copied across and executed locally server-side) - would using that be a more viable alternative?
Without knowing what commands you're intending to run it's hard to take it much further.
EDIT(2)::
If it's just the one command you're running, simply store it in a dedicated file, like 'remote_dir_listing.cmd', and then use psexec with:
psexec \\server -u <user> -p <pass> -c -f remote_dir_listing.cmd
This will force a copy of the local file to the remote side each time you execute it (in case you want to expand it). In this way, you bypass the need for a pause at all - only when psexec has got the pipes open will it run, and once it completes, it closes itself silently.