whizard is hosted by Hepforge, IPPP Durham

Opened 10 years ago

Closed 10 years ago

#674 closed defect (duplicate)

Memory growing during scan

Reported by: kilian Owned by: kilian
Priority: P0 Milestone: v2.2.3
Component: core Version: 2.2.2
Severity: normal Keywords:
Cc:

Description

Reported by Jan Strube:

Memory usage grows during a straightforward parameter scan with integration. Can become a problem if the scan is very fine-grained.

Attachments (1)

memleak.sin (313 bytes) - added by kilian 10 years ago.

Download all attachments as: .zip

Change History (5)

Changed 10 years ago by kilian

Attachment: memleak.sin added

comment:1 Changed 10 years ago by kilian

Status: newassigned

Just trying to reproduce ...

comment:2 Changed 10 years ago by kilian

Reproduced.

IIRC, new processes (runs) are always appended to the process stack, so the linear grow is expected. The process stack is global, not local to the scan iteration. But I have to check this.

I never thought about a number of runs that large. In the present context, it makes sense to delete results that are obtained within a scan iteration. Maybe process/run records should be treated as strictly local ... but this is not that easy to implement.

comment:3 Changed 10 years ago by kilian

I'll make the process (run) stack local to local environments. This will hide their results completely from the global environment, but previously it was only the last scan iteration that was accessible, anyway. There should be arrays of Sindarin objects to catch individual scan results, but that's a different issue.

comment:4 Changed 10 years ago by kilian

Resolution: duplicate
Status: assignedclosed

The main leak should be fixed with r6103. I still see some leakage, but less severe. For now, I close the ticket as duplicate, referring to it in #623.

Note: See TracTickets for help on using tickets.