Re: Re: New Net_Gearman work

On Sun, Apr 4, 2010 at 6:49 AM, Brian Moon <brian@moonspot.net> wrote: >> Hash a jobs unique id to always send it to the same server. Other >> clients do this already. So, I think it is expected. getConnection() >> needs to take an optional unique id from submitTask and do the hashing >> there. A simple modulus hashing would be all that is needed. It would >> need to detect downed gearmand servers and reselect ideally. > > An initial hashing is done in my fork. It does not detect and reselect, but > the lib never did that. So, this is still better than it was before. > >> Convert the library over to the new IANA port. (this is partially done >> in my github fork) > > Ok, all done in my fork. > Sounds good. One thing I noticed with the last release is that the callbacks don't always seem to run, or maybe not at all. I'll try to debug it more next week. I was able to inject the monitor in beginWork(), but all my JOB_START, JOB_FAIL and JOB_COMPLETE never trigger, no errors in the logs either. Regardless of the above, it would be great to get a new release out asap, also a changelog detailing the changes (especially in regard to BC) would be nice. Do you need help with the release process? Till Till