Re: Questions about Tez under the hood

2015-01-30 Thread Fabio C.
Thanks a lot. Now I was trying to figure out what happens if there are different tasks from two different vertexes at the same priority awaiting for resources. If a suitable container becomes available, who is going to get it? I'm having a hard time finding the code about this. I was thinking that

Re: Failed to delete tez scratch data dir - kerberos secured

2015-01-30 Thread Johannes Zillmann
No idea here ? Johannes > On 15 Jan 2015, at 18:07, Johannes Zillmann wrote: > > Hey guys, > > running Tez-0.5.2 with secure impersonation / kerberos. Importing data from > secured hive. > Job is successful. But in the logs i find those warnings: > > -

Re: Failed to delete tez scratch data dir - kerberos secured

2015-01-30 Thread Hitesh Shah
Hi Johannes, Sorry - missed this email earlier. I have seen this a couple of times but have not been able to track down the root cause of this. Would you mind filing a jira for this with the logs from an AM where you observed this? Also, if you have a higher frequency of being able to reproduc

RE: Questions about Tez under the hood

2015-01-30 Thread Bikas Saha
There are many nuances like container affinity etc. but broadly this is what happens. The scheduler loops through all free containers. For each container, at node/rack/any level of locality, it picks the current top priority of tasks that needs to be assigned. For that priority it asks YARN AMR

[ANNOUNCE] Apache Tez 0.6.0.

2015-01-30 Thread Jonathan Eagles
The Apache Tez team is proud to announce the release of Apache Tez version 0.6.0 The Apache Tez project is aimed at creating a framework to build efficient and scalable data processing applications that can be modeled as data flow graphs. This is an feature release that contains the first impleme