I only use oarse grain mode in production, the fine-grained mode can't flexible in any time.
2015-11-24 18:34 GMT+08:00 Rad Gruchalski <[email protected]>: > Indeed, also, keep in mind, there’s a talk here that fine-grained mode > might be removed in Spark 2. > > Kind regards, > Radek Gruchalski > [email protected] <[email protected]> > de.linkedin.com/in/radgruchalski/ > > > *Confidentiality:*This communication is intended for the above-named > person and may be confidential and/or legally privileged. > If it has come to you in error you must take no action based on it, nor > must you copy or show it to anyone; please delete/destroy and inform the > sender immediately. > > On Tuesday, 24 November 2015 at 11:32, 木内満歳 wrote: > > I agree. "coarse grain mode" is "take me to safe side" mode, I think. > Unfortunately, it looks more hard to resolve... > > Anyway, I appreciate your advice. Thanks much ! > > Mitsutoshi > > > 2015-11-24 19:23 GMT+09:00 Rad Gruchalski <[email protected]>: > > Ah, I see, I am experiencing a similar thing with fine-grained where one > of the tasks would stay in staging and fail the whole job but never in > coarse mode. > > Kind regards, > Radek Gruchalski > [email protected] <[email protected]> > de.linkedin.com/in/radgruchalski/ > > > *Confidentiality:*This communication is intended for the above-named > person and may be confidential and/or legally privileged. > If it has come to you in error you must take no action based on it, nor > must you copy or show it to anyone; please delete/destroy and inform the > sender immediately. > > On Tuesday, 24 November 2015 at 10:07, 木内満歳 wrote: > > Hi Rad, > > I've tried both. I've experienced same symptom on both case. > > Thanks, > Mitsutoshi Kiuchi > > > 2015-11-24 17:57 GMT+09:00 Rad Gruchalski <[email protected]>: > > Mitsutoshi, > > Is this in a fine-grained mode? > > Kind regards, > Radek Gruchalski > [email protected] <[email protected]> > de.linkedin.com/in/radgruchalski/ > > > *Confidentiality:*This communication is intended for the above-named > person and may be confidential and/or legally privileged. > If it has come to you in error you must take no action based on it, nor > must you copy or show it to anyone; please delete/destroy and inform the > sender immediately. > > On Tuesday, 24 November 2015 at 07:18, 木内満歳 wrote: > > Hi, Tim > > I've reproduced and taken debug logs(attached). > I cannot understand what is going on, but it seems that the slave is > repeatedly sending ACCEPT message to master. > > Please have your comment. > > Best Regards, > Mitsutoshi Kiuchi > > > 2015-11-24 5:28 GMT+09:00 Tim Chen <[email protected]>: > > Hi Mitsutoshi, > > Can you enable TRACING log on Spark (modify your log4j.properties file)? > > It should have more information on why offers are being rejected, but most > of the time it's due to not enough resources in your cluster to satifsy > launching your Spark job. You can either increase your slave(s) resources > or lower your cpu/memory requirement for your job through configuration. > > Tim > > On Mon, Nov 23, 2015 at 6:30 AM, 木内満歳 <[email protected]> wrote: > > Hi, > > I'm experiencing that some spark task on Mesos 0.25 occasionally won't > start. > Please tell some advice how to see more detail against it. > > Here is the slave log about bad task > > Nov 23 08:54:26 mesos-s2 mesos-slave[18499]: I1123 08:54:26.677291 18516 > slave.cpp:2379] Got registration for executor > '235498ca-6603-4cfe-bfc7-94005bb235fb-S5' of framework > 235498ca-6603-4cfe-bfc7-94005bb235fb-1442 from executor(1)@ > 10.130.91.16:60295 > Nov 23 08:54:26 mesos-s2 mesos-slave[18499]: I1123 08:54:26.679875 18516 > slave.cpp:1760] Sending queued task '0' to executor > '235498ca-6603-4cfe-bfc7-94005bb235fb-S5' of framework > 235498ca-6603-4cfe-bfc7-94005bb235fb-1442 > (no more log about this task) > > When task succeed to run, slave log shows like that. > > Nov 23 08:44:39 al-mesos-s3 mesos-slave[8644]: I1123 08:44:39.637285 8658 > slave.cpp:2379] Got registration for executor > '235498ca-6603-4cfe-bfc7-94005bb235fb-S6' of framework > 235498ca-6603-4cfe-bfc7-94005bb235fb-1437 from executor(1)@ > 10.130.98.65:52273 > Nov 23 08:44:39 al-mesos-s3 mesos-slave[8644]: I1123 08:44:39.639233 8658 > slave.cpp:1760] Sending queued task '6' to executor > '235498ca-6603-4cfe-bfc7-94005bb235fb-S6' of framework > 235498ca-6603-4cfe-bfc7-94005bb235fb-1437 > Nov 23 08:44:42 al-mesos-s3 mesos-slave[8644]: I1123 08:44:42.608182 8658 > slave.cpp:2717] Handling status update TASK_RUNNING (UUID: > ff5a2278-0753-4541-bd33-a55f3a09fb69) for task 6 of framework > 235498ca-6603-4cfe-bfc7-94005bb235fb-1437 from executor(1)@ > 10.130.98.65:52273 > Nov 23 08:44:42 al-mesos-s3 mesos-slave[8644]: I1123 08:44:42.612318 8658 > status_update_manager.cpp:322] Received status update TASK_RUNNING (UUID: > ff5a2278-0753-4541-bd33-a55f3a09fb69) for task 6 of framework > 235498ca-6603-4cfe-bfc7-94005bb235fb-1437 > > Any advice is welcome. > > Best Regards, > Mitsutoshi Kiuchi > > > > > Attachments: > - log.driverStdErr.gz > - log.mesosMaster.gz > - log.mesosSlave.gz > > > > > > > -- Deshi Xiao Twitter: xds2000 E-mail: xiaods(AT)gmail.com
