Re: Session timeout when performing an action
Re: Session timeout when performing an action
- Subject: Re: Session timeout when performing an action
- From: Klaus Berkling <email@hidden>
- Date: Tue, 15 Oct 2013 13:54:55 -0700
I'm somewhat puzzled. Do you have one of these long requests going on when you get the improper session timeout error?
Or is there a lengthy request that was aborted by the user/client? Because that doesn't stop the database query. Eventually you'll get a broken pipe error in the log long, long after.
So if there is a long, aborted, query going on and the action that a user clicked on is a database request, that second database request will wait for the first query to finish.
Just wondering, have you ever tinkered with WOAllowConcurrentRequestHandling in the properties file?
On Oct 15, 2013, at 8:40 AM, Chuck Hill <email@hidden> wrote:
>
>
> On 2013-10-15 2:52 AM, "Schoenenberger Dominique"
> <email@hidden> wrote:
>
>> Thanks a lot, that's informative.
>>
>> Dead time: 30
>> I have only 1 active request over the 13 instances.
>> Yes 20 min is really too much: We have some request which can take up to
>> 1-2 min.
>
> You want to use a WOLongResponsePage (or Wonder version) for these. And
> probably a separate EOF stack if they are working with EOs.
>
> Chuck
>
>
>> In the apache log, I saw http error 500.
>>
>> I found the source of the "No error" displayed in the browser: It comes
>> from the adaptor when the response is null and error has not been set
>> (see transaction.c)
>>
>> Dominique
>>
>> On Oct 15, 2013, at 10:54 AM, Alexander Spohr <email@hidden> wrote:
>>
>>> Did you check /WebObjects/WOAdaptorInfo?
>>> What is the dead time?
>>> Do you see a lot of active requests?
>>>
>>>
>>> Am 14.10.2013 um 15:26 schrieb Schoenenberger Dominique
>>> <email@hidden>:
>>>
>>>> Thanks,
>>>>
>>>> My parameters are (bigger):
>>>>
>>>> HTTP Adaptor Settings:
>>>> Load Balancing scheme: Round Robin
>>>> Send Timeout: 1200
>>>> Receive Timeout: 1200
>>>
>>> Do you have requests that take longer than 20 minutes?
>>>
>>>
>>>> The page seems related to an http error 500.
>>>
>>> Is it a 500 or another 5xx?
>>>
>>>
>>>> Is any way to avoid the adaptor to give the request to another
>>>> instance if one instance is too slow to respond because this seems no
>>>> helping because at the end it causes a "Session timeout" error.
>>>
>>> This would just hide your problem. You need to fix it.
>>>
>>>
>>>> I need also to say that when the user get a untimely "Session
>>>> Timeout", there is not waiting time (We got the session timeout just
>>>> after the click on a button). I would expect some retrying before
>>>> returning an Session timeout error if it is due to a instance not
>>>> responding.
>>>
>>> The apache has already decided that your instance is dead because it
>>> did not answer in time. So it just kicks your request to the next
>>> instance.
>>>
>>>
>>
>>
>> _______________________________________________
>> Do not post admin requests to the list. They will be ignored.
>> Webobjects-dev mailing list (email@hidden)
>> Help/Unsubscribe/Update your Subscription:
>> age.net
>>
>> This email sent to email@hidden
>
>
> _______________________________________________
> Do not post admin requests to the list. They will be ignored.
> Webobjects-dev mailing list (email@hidden)
> Help/Unsubscribe/Update your Subscription:
>
> This email sent to email@hidden
kib
Klaus Berkling
www.berkling.us | @kiberkli
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Webobjects-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
This email sent to email@hidden