[wpkg-users] Global max execution time hardcoded?

[le dahut]

Hi,

Maybe another solution : Can WPKG-GP let users login / present login 
screen before it has finished its work ?
Is there such an option to define ?

This way, administrators could open a session and see what is blocking.


  Klaas


20/03/2013 20:43, Rainer Meier :
> Hi Marco,
>
>
> On 20.03.2013 19:06, Marco Gaiarin wrote:
>>
>> In my setup, switching to 7/WPKG-GP, i need to impose a rather
>> draconian ''max execution time'' (man 15 minutes) to wpkg run.
>>
>> After looking in docs and in the code, seems to me that the 'one hour'
>> limit is hardcoded, but there's two places, eg:
>
> As you already found out all executions of external applications are
> done via the exec() function. If you like to change the timeout then
> feel free to adjust it, it's open source. I currently don't see a reason
> to configure it globally as you can also set individual execution times
> on individual commands and nothing else you will change if you change
> the exec() function.
>
>
> There is no GLOBAL execution timeout for a complete WPKG run (it can
> basically run for hourws). But only a global timeout for individual
> commands which by default is set to 1h unless timeout is specified for
> an individual command.
>
> The purpose of 1h timeout per command is only to assure that WPKG will
> not wait forever on commands which block (e.g. cmd scripts which have a
> "pause" in it or simply installers which won't proceed without user
> input). So it's set reasonably high so even large programs will usually
> finish within the timeout.
> If you know that a specific command is taking much less time you can
> also lower the timeout but it will have an impact only if this command
> will exceed the timeout. WPKG of course does not wait until timeout
> expires if the command terminates earlier.
>
> Please note that lowering the timeout for an individual account still
> does not allow you to make any easy estimate on the total WPKG run time.
> Run time depends on number of commands in your packages. Package
> installation will actually be marked as failed if timeout occurs. So I
> personally really don't see a reason to lower the timeout unless you're
> keen on seeing more errors during deployment. Perhaps it makes sense to
> lower the timeout for specific commands but this is what the timeout
> attribute is for.
>
> br,
> Rainer