uima-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Eddie Epstein <eaepst...@gmail.com>
Subject Re: Error on getMeta call to remote service
Date Wed, 16 Dec 2009 12:24:28 GMT
ActiveMQ has an optional broker plugin that adjusts time-to-live such that only
the broker and service machines have to be synchronized. We couldn't get this
to work on AMQ 4.1.1, but assume it has been fixed in 5.2. Feel free to create
an issue and we can use it to at least fix documentation, if not more.


On Wed, Dec 16, 2009 at 5:02 AM, Prokopis Prokopidis <prokopis@ilsp.gr> wrote:
>> On Tue, Dec 15, 2009 at 3:52 PM, Prokopis Prokopidis <prok...@gmail.com>
>> wrote:
>> > If it's any help, when calling the service from the
>> > machine on which it has been deployed, or when reversing
>> > the service-client roles between the two machines,
>> > everything works as in the README file.
>> Hmmm, smells like a clock synchronization problem. In order to avoid
>> undeliverable messages building up on a broker, timeout settings
>> are used to create time-to-live properties in the JMS messages.
>> ActiveMQ implements time-to-live based on the client and server
>> machine clocks.
>> Eddie
> Thank you! Following your advice and after synchronizing the clocks on both
> machines (via ntpd) I managed to call the service as expected.
> Since I am no ActiveMQ expert, I'm now left wondering whether there is a way
> to ensure that similar problems do not come up if a real-life service is
> accessed by clients with unsynchronized clocks (on which clients, of course,
>  I will not be able to synchronize the clock).
> Is this an issue which can be resolved in the client or server side code and
> in the example of the AS distribution as well? Should I create an issue in
> Best,
> Prokopis

View raw message