On 23/11/13 15:22, Ben Noordhuis wrote:
> I can't really reproduce that.  Below is what IMO the test case
> reasonably should look like and that has RSS hovering around 60 MB
> with --max_old_space_size=64.  Tested with HEAD of v0.10 and master on
> a x86_64 FC20 system.
>
>

I tried this on both 0.8 (with process.nextTick instead of setImmediate)
and 0.10 and it does not show a leak or excessive memory usage.

However - this would have been quite different to what was happening in
Haraka with the bug that we had there; we would have been calling
fs.unwatchFile(name), fs.watchFile(name) per line of data or per buffer
which could have been potentially thousands of watchers per second in
the same stack.  I was also running 0.8 on all of the machines that were
exhibiting this problem.

What I think happened here is that the fs.watch/unwatchFile functions
saturated the thread pool to the point that all the threads were busy
and it couldn't keep up which caused the RSS to grow with all of the
in-flight requests.

Regards,
Steve.

-- 
-- 
Job Board: http://jobs.nodejs.org/
Posting guidelines: 
https://github.com/joyent/node/wiki/Mailing-List-Posting-Guidelines
You received this message because you are subscribed to the Google
Groups "nodejs" group.
To post to this group, send email to [email protected]
To unsubscribe from this group, send email to
[email protected]
For more options, visit this group at
http://groups.google.com/group/nodejs?hl=en?hl=en

--- 
You received this message because you are subscribed to the Google Groups 
"nodejs" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
For more options, visit https://groups.google.com/groups/opt_out.

Reply via email to