Help on long running tasks

[5u623l20]

Currently we can exceed the 120m time limit if either of the following conditions are met:

• Persistent Worker
• Compute Service
• Compute Credits

Is there any way to add a single logic to a task that it will be executed only if either of the above is true. I know it also creates a race condition as what if two of the conditions are met and where is the task supposed to run?

I can create three different tasks and keep it in the file but the problem is the URL of the artifacts will change hence it is difficult to integrate in a dashboard.

[fkorotkov]

Could you please elaborate your use case a bit? If the problem only in the artifacts URLs then you might be interested in aliases which can be used in URLs to artifacts. If all the tasks will have the same alias then the URL prefix will be the same for them.

[5u623l20]

We have couple of tasks on different architectures. The build times are fine which are under 90m actually but the problem is with the tests. Those takes long hours around 6-8 hours for non-x86 ARCHS which we want to run in any of three mentioned above.

alias looks like a good option. I have to add more logic to task depends and only_ifs because in case more than two options are defined two same tasks are going to be executed and writing in the same artifact location. I am not sure about the behavior on what happens when there are same filenames of same artifacts although it says it's saved under a zip file but I will have to check in details.