Bug 1700915

Oops, I forgot the link to the source code for the parameter:
https://git.beaker-project.org/cgit/restraint/tree/src/task.c#n328

Looking at the flow, it makes sense to me though.
1. You define the max time in the task metadata
2. In case you have RSTRNT_MAX_TIME defined you overwrite that value
3. Value of max time is extracted to RSTRNT_MAXTIME

I'm not sure if we are making any changes to max time between the steps 2 and 3 but in case we need to that in the future, we won't cause any breaking. But I understand the point and I would actually be for deprecating RSTRNT_MAX_TIME in favor for KILLTIMEOVERRIDE so there's no confusion.

Just to let you know, the RSTRNT_ prefix was used to keep from possibly contaminating the environment variables..

The inconsistency is confusing, though, for writing a task vs running it.

When writing a task, you might check it like so:

  if [ $RSTRNT_MAXTIME -lt $THRESHOLD ]; then
    ...
  fi

And then you schedule a Beaker job to run your task and have to set a different name:

  <params>
    <param name="RSTRNT_MAX_TIME" value="100"/>
  </params>

Trying to remember which one takes the underscore and which one does not is a recipe for a headache.

(In reply to Jeff Bastian from comment #4)
> The inconsistency is confusing, though, for writing a task vs running it.
> 
> When writing a task, you might check it like so:
> 
>   if [ $RSTRNT_MAXTIME -lt $THRESHOLD ]; then
>     ...
>   fi
> 
> And then you schedule a Beaker job to run your task and have to set a
> different name:
> 
>   <params>
>     <param name="RSTRNT_MAX_TIME" value="100"/>
>   </params>
> 
> Trying to remember which one takes the underscore and which one does not is
> a recipe for a headache.

I get that Jeff,  I was suggesting we support both and only document one.  I believe Tomas suggestions was to document the non name-spaced KILLTIMEOVERRIDE.

Restraint 0.1.40 has been released.