Open
Bug 941882
Opened 11 years ago
Updated 2 years ago
Programmatically get the memory usage of a worker
Categories
(Toolkit :: Async Tooling, defect)
Toolkit
Async Tooling
Tracking
()
NEW
People
(Reporter: Yoric, Unassigned)
Details
No description provided.
Comment 1•11 years ago
|
||
What do you mean by this?
Updated•11 years ago
|
Whiteboard: [Memshrink] → [MemShrink]
Reporter | ||
Comment 2•11 years ago
|
||
We would like to report through Telemetry the memory usage of specific workers. This requires an API that lets us get the information. I can live with doing this either from the worker itself or from the main thread.
Comment 3•11 years ago
|
||
This is hard for two reasons. First, we have lots of reporters for different parts of the system. But I'm trying very hard to avoid treating any of them as special. E.g. about:memory just iterates through and displays all of them. We actually used to pull out some of them specially -- telemetry did exactly this -- and then some of those reporters went away, and the implicit dependence was broken, and nobody noticed for months. This is bad, and is why I introduced "distinguished amounts" in bug 913260 to fix this exact situation. What you're asking goes against this. Maybe there is a way to do it nicely, but... Second, telemetry measurements have to be fast. We've had multiple issues in the past with memory measurements that were reported by telemetry that were too slow, and caused hangs, and so were removed. Getting the memory consumption of a worker requires iterating over every live thing in the worker's GC heap, and measuring all the malloc'd blocks that hang off those things. This is not something that is appropriate for doing once per minute (or however telemetry runs).
Updated•11 years ago
|
Whiteboard: [MemShrink]
Updated•2 years ago
|
Severity: normal → S3
You need to log in
before you can comment on or make changes to this bug.
Description
•