Hey Thanks for the reply Murvin: answers to your config questions below

I should probably mention first that this is a small / low pressure server
(a small server with 256MB RAM is plenty to handle our traffic, which
realistically never amounts to more than than 2-3requests/sec when very
busy)

- Running with default memory config
- We're not running with socket.io or express or any other package with
production v. default configuration, so I'd have to say this probably
doesn't apply (unless I am misunderstanding your question?)
- We run the node process without any system flags - as I understand them
both -nouse-idle-notification & -expose-gc can be used to optimize garbage
collection when GC is a problem but I don't have any reasons to think that
would be the case here
- The server makes plenty of remote service calls if by service call you
mean http requests to 3rd party servers. What did you have in mind there?
- Memory usage and cpu seemed normal before the glitch aka 30-40MB heap,
CPU in the low single digits (note that the server is still serving
requests after missing the probe request from monit - it simply seem to be
*missing* a request once in a while)

On Tue, Sep 25, 2012 at 8:48 AM, Murvin Lai <[email protected]> wrote:

> few things you may want to check.
>
> - are you running with default 1GB memory or increase to ~2GB?
> - default or production mode?
> - I usually run with this flag:  node -nouse-idle-notification -expose-gc
>   myHttpServer.js
> - does your server make any remote service call?
>
> and what is the CPU / memory usage during the glitch?
>
>
> On Mon, Sep 24, 2012 at 4:54 PM, Hugo <[email protected]> wrote:
>
>> Confirmed :( I have been running almost 2 weeks with node 6.18 without a
>> glitch. With 0.8.9 I was seeing 2-3 issues per day at the minimum
>>
>> Unfortunately I don't have much of a diagnosis, I found out the issue
>> because I am monitoring the server with monit <http://mmonit.com> (it
>> tests the server with a basic http request that does nothing other than
>> acknowledging with a 200 code)
>>
>> For every failure, artifacts of the monit request was missing from the
>> logs - which leads me to believe that the server never acknowledged the
>> request(?)
>>
>> Note that the version of node is the only factor that was changed between
>> tests (same code, same server, same config for all components)
>>
>>
>> Has anyone experience anything of a similar nature?
>>
>> --
>> 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
>>
>
>  --
> 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
>

-- 
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

Reply via email to