i got an error 500 while executing a slow SQL query (fetched some 100k
rows on a join over a dozen tables due to poor coding, which took 75
seconds to complete).
The app completed the slow query w/o errors on thin or mongrel.
So i wondered if webroar has a timeout defined after which it throws a
500? If so, is there a way to configure this?
(using version 0.3.2 before today's build)
Thanks in advance
--hl
--
To unsubscribe, reply using "remove me" as the subject.
.. and excuse my (by now well documented) sluggishness to examine the
code by myself.
On Apr 6, 2:19 pm, Nikunj Limbaseeya <nikunj.limbase...@webroar.in>
wrote:
> [..]
> - After 60 seconds it sends first PING signal and waits for 15 sec to get a reply.
does Head awaits the worker process to complete, respectively in which
case the worker replies to a ping?
e.g. the app (worker) sends out several hundreds of emails via a
remote SMTP connection, (and this could take some time. No, it's not
spam.). Does the worker process reply to the head's ping, while it's
busy processing - or just when idle?
--hl
(1.25 thumbs up for WebRoaR!)
--
Hi,
sorry for coming back to this issue:
The git repository contains 0.3.1 as the latest version. Actually i do
have version 0.3.2 running.
$ webroar --version
WebROaR-0.3.2
I can't find WR_WKR_IDLE_TIME anywhere in the sources of 0.3.1.
This raises two questions:
1. Where did i get version 0.3.2 from? :-)
2. How to handle timeouts in version 0.3.1?
Thanks in advance
--hl
> Harm,
>
> Thank you for trying out WebROaR and posting the issue on the group.
>
> We have implemented a mechanism to identify the worker which hangs while
> processing a request. As per this logic if a worker is unable to process the
> request within 90 seconds, it sends 500 error response to end user.
>
> Summarizing the scenario.
>
> - Head waits for 60 seconds to get processed request back from worker.
> - After 60 seconds it sends first PING signal and waits for 15 seconds to
> get a reply.
> - After 15 seconds it sends second PING signal and again waits for 15
> seconds to get some reply.
> - If worker is not responding during this time interval, the head assumes
> the worker is in a unstable state and unable to process further requests.
>
> We have defined values for worker idle time (WR_WKR_IDLE_TIME), ping wait
> time (WR_PING_WAIT_TIME) and number of ping trials (WR_PING_TRIALS) in
> 'wr_config.h' file.
>
> For time being to solve your issue and run your application well you can
> increase these values and recompile the code.
>
> We have also planned to make those values configurable in a future release.
>
> Thanks,
> Nikunj
>