On Sun, May 16, 2010 at 2:48 PM, Supun Kamburugamuva <supu...@gmail.com>wrote:
> I think Even for a production scenario setting this to 1 hour is too much. > Normally if a sever crashes it doesn't take 1 hour to restart it. Also from > synapse point of view it is always better to try the failed endpoint more > frequently than waiting for 1 hour dropping messages. > > +1 As per reading i have done for DLC implementation idea of having a redelivery mechanism is, in most of the cases a messages may fail due to higher load like minor issues at the endpoint.in that scenarios a simple re attempt will solve the issue. So suspeding a endpoint for an hour for that kind of senarios is a huge waste. There is always a trade off senario between cost of attepmting a message delivery to a failed endpoint vs cost of not enabling a working endpoint. I prefer trying the failed endpoint more frequently than disabling it for an hour. > Thanks, > Supun.. > > > On Sun, May 16, 2010 at 2:23 PM, Hiranya Jayathilaka <hiranya...@gmail.com > > wrote: > >> On Sun, May 16, 2010 at 12:17 PM, Miyuru <miyur...@gmail.com> wrote: >> >>> +1 >>> it will be more convenient and more closer to real user requirements. >>> >> >> Not exactly. IMO 1 hour is suitable for most production scenarios. In a >> real system, if an endpoint gets suspended, that usually means something has >> terribly gone wrong (a backend server may have gone down, a connection may >> have dropped etc). Recovering from such an error could take a long time. I >> think the default suspend duration is set to 1 hour considering these >> issues. But I agree that during development time this could be major pain. >> In development phase endpoints may get suspended all the time due to various >> reasons. In that sense it is beneficial to have a much shorter default >> suspend duration for endpoints. >> >> Hence I'm actually +0 on this right now. >> >> Thanks, >> Hiranya >> >> >>> 1 hour is tooooo much. I think 1min is more appropriate. >>> >>> >>> On Sun, May 16, 2010 at 10:17 AM, Supun Kamburugamuva <supu...@gmail.com >>> > wrote: >>> >>>> At the moment default endpoint suspension time is set to 60 * 60 * 1000 >>>> miliseconds. This means with simple scenarios if a endpoint gets suspended >>>> user has to restart synapse. How about reducing this default value to lets >>>> say 30 secs or 1 min? >>>> >>>> Thanks, >>>> Supun.. >>>> >>>> -- >>>> Software Engineer, WSO2 Inc >>>> http://wso2.org >>>> supunk.blogspot.com >>>> >>>> >>>> >>> >>> >>> -- >>> Miyuru Daminda Wanninayaka >>> Software Engineer - WSO2 Inc. >>> >> >> >> >> -- >> Hiranya Jayathilaka >> Senior Software Engineer; >> >> WSO2 Inc.; http://wso2.org >> E-mail: hira...@wso2.com; Mobile: +94 77 633 3491 >> Blog: http://techfeast-hiranya.blogspot.com >> > > > > -- > Software Engineer, WSO2 Inc > http://wso2.org > supunk.blogspot.com > > > -- Charith Dhanushka Wickramarachchi http://charithwiki.blogspot.com/