I've tried to execute it at the VM and I think the script won't execute gunicorn because we're using the prod server venv path

[root@vm development]# source /srv/www/htdocs/AutosubmitAPI/venv27/bin/activate
bash: /srv/www/htdocs/AutosubmitAPI/venv27/bin/activate: No such file or directory

Instead it should be the following:

[root@vm development]# source /var/www/html/autosubmitapi/venv27/bin/activate
bash: /srv/www/htdocs/AutosubmitAPI/venv27/bin/activate: No such file or directory

I've already made the changes in the VM's script

Thanks @cgutierr, you mean in the Dev. VM right? I tested this one in the production machine.

@jberlin I guess you didn't test the script in the VM?

@mcastril yes, all of this has been done in the dev VM.

Hi @mcastril , no I did not tested it. I thought the virtual dev env was set elsewhere and not in the script

we can pass the path to the virtual environment as parameter

I also added the kill gunicorn lines to the VM script, they weren't there.

I will add the parameter with the virtual env path

new call will be as follows:

./update_launch_autosubmit_API.sh <path_to_env> | --update [OPT]

examples:

./update_launch_autosubmit_API.sh /var/www/html/autosubmitapi/venv27/bin/activate
./update_launch_autosubmit_API.sh /var/www/html/autosubmitapi/venv27/bin/activate --update

Can you set the path in an environment variable (in the .bashrc) and use it in the script? And allow the script to override the variable.

As it is getting more complex, you can use Bash getopts, and define the following options:

-e  to set the Python virtual env path
-u  to do the update

The Python ve default path would be the one defined in the environment variable defined in the .bashrc.

This will make it easier to run the script out of the crontab.

For example:

PVE=$PVE_PATH   # PVE_PATH is an env var defined in the .bashrc
UPDATE=false

while getopts ":e:u" opt; do
  case "$opt" in
    e)  PVE=$OPTARG
      ;;
    u)  UPDATE=true
      ;;
   esac

if [ "$UPDATE" = true ]; then
....

I didn't check the code...

crontab en la VM:

@reboot /home/rocky/development/update_launch_AS_API.sh var/www/html/autosubmitapi/venv27/bin/activate

crontab en prod:

@reboot /home/webadmin/update_launch_AS_API.sh /srv/www/htdocs/AutosubmitAPI/venv27/bin/activate

I tested only in dev environment

Hi @mcastril , I will have a look to your idea, but isnt .bashrc user specific ? I think it can cause more dependence on the user. but I will have a look anyways.

Yes, it's user-specific, but we are all using root to launch the API (it's in root's crontab and you can only kill root's processes if you are root).

So putting the env var in root's .bashrc (webadmin in production) in the VM should work.

This is to not have to remember the path everytime (that is different in both systems) to re-launch the system in case of emergency.

Hi @mcastril , I applied your idea in dev environment and it works

I applied the changes to all environments, I did not test in production to avoid causing disruptions

Perfect @jberlin

Tested in production now. It works.