The upgrade procedure or the command unistart fail in case one of the user services can't be started.
This can disrupt production as the node will not start anymore after launching the autopatch procedure.
It can be reproduced on purpose on our end when we delete one of the user services manually ( command sc del) or if we add a non-existing user service in %U_TMP_PATH%\user_services.txt
First, stop the node via the command "unistop all".
Then edit the file %U_TMP_PATH%\user_services.txt and add a new line with a non-existing user service, the unistart script will fail and will not start the node.
Another way to reproduce is to delete an user service after having stopped the node.
C:\Automic\DUAS\TST600_NB120285\bin>sc delete TST600_NB120285_univ_uxpro500e
[SC] DeleteService SUCCESS
Instance startup procedure
checking version file consistent . . . .
Version file consistent with binary.
Checking data files status . . . .
Data files status ok
Starting users services . . . .
Starting service "TST600_NB120285_univ_administrator"
The Univer$e TST600_NB120285 user administrator service is starting.
The Univer$e TST600_NB120285 user administrator service was started successfully
Starting service "TST600_NB120285_univ_uxpro500e"
The service name is invalid.
A Dollar Universe Windows User Service had been deleted manually on the Windows Services, by default, "unistart all" command will try to start every Windows Service that had been stopped previously via "unistop" command.
Submission Accounts must always be deleted via UVC or the command line ( uxdlt user user=name_of_user).
In case the node cannot be started, delete the file %U_TMP_PATH%\user_services.txt and restart the node.