Re: [jetty-dev] HttpClient version 7.0.1.v20091125: few cases that asyn

[Jesse McConnell <jesse.mcconnell@xxxxxxxxx>]

7.0.2.v20100331 was also released last week so it might be worth
verifying the issues still exist..

cheers!
jesse

--
jesse mcconnell
jesse.mcconnell@xxxxxxxxx



On Sun, Apr 4, 2010 at 23:08, Paul Yang <pccyang@xxxxxxxxx> wrote:
> Already did.  :)    Bug 307457
>
> -Paul
>
> On Sun, Apr 4, 2010 at 1:24 AM, Simone Bordet <simone.bordet@xxxxxxxxx>
> wrote:
>>
>> Hi,
>>
>> On Wed, Mar 24, 2010 at 00:30, Paul Yang <pccyang@xxxxxxxxx> wrote:
>> > Hi,
>> >
>> > Under the following cases, some asynchronous exchanges won't come back
>> > (actually some are never sent). That means neither onResponseComplete(),
>> > onConnectionFailed(), onException() or onExpire() is invoked.
>> >
>> > 1. connection failed and HttpDestination::_idle queue has more than
>> > allowed
>> > MaxConnectionsPerAddress pending exchanges.
>> > Under this case, the current implementation just calls
>> > onConnectionFailed()
>> > of the first exchange from _idle queue. The rest become orphan and never
>> > get
>> > send.
>> > Possible fix could be; calling onConnectionFailed() for all exchanges in
>> > _idle queue, or just calls onConnectionFailed() of the first exchange
>> > from
>> > _idle queue then try to start a new connection if the _idle is still not
>> > empty
>> >
>> > 2. HttpConnection::send failed
>> > When this happened, connection was neither in the idle timer or timeout
>> > timer but still in HttpDestination::_connection queue, so this
>> > connection
>> > never got a chance to be removed and it occupied one slot from
>> > MaxConnectionsPerAddress. These connections will gradually eat up all
>> > the
>> > maximum allowed connections per address and no more exchanges can be
>> > send.
>> >
>> >
>> > It might be easier to see it in action. The attached files contain a
>> > quickly
>> > whipped up test program (to reproduce the problem) and patch against
>> > version
>> > 7.0.1.v20091125.
>>
>> Thanks, can you please raise a Bugzilla for these issues ?
>>
>> Simon
>> --
>> http://bordet.blogspot.com
>> ---
>> Finally, no matter how good the architecture and design are,
>> to deliver bug-free software with optimal performance and reliability,
>> the implementation technique must be flawless.   Victoria Livschitz
>> _______________________________________________
>> jetty-dev mailing list
>> jetty-dev@xxxxxxxxxxx
>> https://dev.eclipse.org/mailman/listinfo/jetty-dev
>
>
> _______________________________________________
> jetty-dev mailing list
> jetty-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/jetty-dev
>
>