Carey,

Some follow up on this.

I wrote a test with async servlet handling accessing sessions, and it
passes:  the request count is as it should be.

I need more details from you:
+ do you wrap your requests
+ do you have multiple simultaneous requests accessing the same session
+ what SessionCache implementation do you use, and what configuration do
you have on it
+ what is the max inactive interval on your sessions
+ how does your async handling terminate, is it with dispatch() or
complete()
+ do you do any forwarding between contexts
+ does your application call Request.changeSessionId()
+ does your application use form authentication

Jan

On Thu, 13 Dec 2018 at 13:38, Jan Bartel <[email protected]> wrote:

> Hi Carey,
>
> So sessions will only be written to the store (jdbc) when the request
> count drops to zero. If the count remains above 0, then that is why they
> are not written nor expired.
>
> I will take a particularly close look at the Async handling myself, but if
> you came up with a repro test case that does some simple async handling
> that would be great .
>
> Jan
>
> On Thu., 13 Dec. 2018, 13:24 Carey Evans <[email protected] wrote:
>
>> Hi.
>>
>> We're trying to use Jetty 9.4.14 with an application that uses a lot of
>> sessions, and we're having trouble with the DefaultSessionCache not
>> deleting some old sessions, until we ended up with 6.7 GB consumed by
>> 393216 Session objects in the heap after our load test. Looking at the
>> sessions that hadn't expired after a few hours, I see that each of them has
>> _requests set to 1, when it looks like they should all be 0.
>>
>> We're using a JDBC session store, and a lot of sessions didn't seem to
>> make it to the database table, although the ones that did were expired and
>> removed as expected.
>>
>> This is a fairly large application with a lot of proprietary code, so
>> unfortunately I don't have a way to reproduce this that I can share at the
>> moment. We do a fair bit of asynchronous servlet processing, though, which
>> makes me suspicious. Where can I start to look for a solution to this?
>>
>>
>>
>> --
>> Laissez lire, et laissez danser; ces deux amusements ne feront jamais de
>> mal au monde.
>> _______________________________________________
>> jetty-users mailing list
>> [email protected]
>> To change your delivery options, retrieve your password, or unsubscribe
>> from this list, visit
>> https://www.eclipse.org/mailman/listinfo/jetty-users
>
>

-- 
Jan Bartel <[email protected]>
www.webtide.com
*Expert assistance from the creators of Jetty and CometD*
_______________________________________________
jetty-users mailing list
[email protected]
To change your delivery options, retrieve your password, or unsubscribe from 
this list, visit
https://www.eclipse.org/mailman/listinfo/jetty-users

Reply via email to