Error with Event Type when Testing Random Day from Google Trace

[duaa h]

I want to run simulation using google cluster dataset. But I don't want to run it from first day, I want to choose any day to test. Task event type causes a problem. for example, if a task occurs first time with kill event (it submitted in previous day), this will cause an error. How can I fix this error?   Anyone could help me please

[john wilkes]

You'd need to calculate the set of active tasks (etc) by reading the part of the trace before your start window.  That's the only way to avoid surprises like the one you encountered.

On Sun, Jun 29, 2025, 10:45 AM duaa h <d.hum...@gmail.com> wrote:

I want to run simulation using google cluster dataset. But I don't want to run it from first day, I want to choose any day to test. Task event type causes a problem. for example, if a task occurs first time with kill event (it submitted in previous day), this will cause an error. How can I fix this error?   Anyone could help me please

--
You received this message because you are subscribed to the "Google cluster data - discussions" group. To post to this group, send email to googlecluste...@googlegroups.com. To unsubscribe from this group, send email to googleclusterdata-...@googlegroups.com. For more options, visit this group at https://groups.google.com/d/forum/googleclusterdata-discuss?hl=en-US.
---
You received this message because you are subscribed to the Google Groups "Google cluster data - discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to googleclusterdata-...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/googleclusterdata-discuss/e333fee8-5515-475f-ac6b-1e9c5f10b8c8n%40googlegroups.com.

[duaa h]

Could you explain more please. I need to read only tasks that submitted in my testing file, otherwise ignored.

[john wilkes]

You need to know which tasks are running in the system at your simulation start point (at least, the ones that you care about).   The only way to do that is to simulate all the start/stop events up to that point.  Or you can make your software assume that any task-kill event for a task that's not previously been seen is for a task that was running at the start of your window (i.e., the "error" is with your software, not the trace :-)).

To view this discussion visit https://groups.google.com/d/msgid/googleclusterdata-discuss/b3dfaf38-605c-4c0e-bbf9-1342f72735a7n%40googlegroups.com.

[duaa h]

Thank you. yes, the problem is from the software. I am using Cloudsim plus framework and I don't know the exact method to modify.

[john wilkes]

That I cannot help you with.  Perhaps somebody else in the group can, though?

To view this discussion visit https://groups.google.com/d/msgid/googleclusterdata-discuss/9d9d1c48-40ee-4322-a420-be6d1f1fe57bn%40googlegroups.com.

[duaa h]

Thank you so much for your interest and trying to help.