Re: Resource under-utilization when using RocksDb state backend [SOLVED]

2017-02-16 Thread Clifford Resnick
Hi Vinay, We found that our problems were not with RocksDb, but rather what we were throwing at it. We were working with more complex data types (e.g. Collections) and found that nearly 80% of the time was spent in serialization, so optimizing that helped a lot. But if your state is more primit

Why are externalized checkpoints deleted on Job Manager exit?

2016-11-02 Thread Clifford Resnick
Testing externalized checkpoints in a YARN-based cluster, configured with: env.getCheckpointConfig.enableExternalizedCheckpoints(ExternalizedCheckpointCleanup.RETAIN_ON_CANCELLATION); I can confirm that checkpoint is retained between cancelled jobs, however it’s deleted when the Job Manager sess

Re: Error using S3a State Backend: Window Operators sending directory instead of fully qualified file?

2016-07-21 Thread Clifford Resnick
CDH5. I'll verify and submit the original PR tomorrow morning EST. On Jul 22, 2016 12:26 AM, Clifford Resnick wrote: I have a fix and test for a recursive HDFSCopyToLocal. I also added similar code to Yarn application staging. However, even though all files and resources now copy correctly,

Re: Error using S3a State Backend: Window Operators sending directory instead of fully qualified file?

2016-07-21 Thread Clifford Resnick
coming 1.1 RC (probably end of this week or beginning of next). Feel free to ping me if you need any feed back or pointers. – Ufuk On Mon, Jul 18, 2016 at 9:52 PM, Clifford Resnick wrote: > In 1.1, AbstractYarnClusterDescriptor pushes contents of flink/lib (lo

Re: Error using S3a State Backend: Window Operators sending directory instead of fully qualified file?

2016-07-18 Thread Clifford Resnick
ve File copy in Flink (in HDFSCopyToLocal) and contribute this via a pull request. – Ufuk On Mon, Jul 18, 2016 at 7:22 PM, Clifford Resnick wrote: > Hi Ufuk, > > My mail was down, so I missed this response. Thanks for that. > > On

Re: Error using S3a State Backend: Window Operators sending directory instead of fully qualified file?

2016-07-18 Thread Clifford Resnick
, Jul 16, 2016 at 6:26 PM, Clifford Resnick wrote: > Using Flink 1.1-SNAPSHOT, Hadoop-aws 2.6.4 > > > > The error I’m getting is : > > > > 11:05:44,425 ERROR org.apache.flink.streaming.runtime.tasks.StreamTask > - Caught

Re: Error using S3a State Backend: Window Operators sending directory instead of fully qualified file?

2016-07-18 Thread Clifford Resnick
ion of Hadoop and Flink, and was there anything you did other than configure core-site.xml? -Cliff From: Clifford Resnick Reply-To: "user@flink.apache.org" Date: Saturday, July 16, 2016 at 12:26 PM To: "user@flink.apache.org" Subject: Error using S3a State Backend: Window O

Error using S3a State Backend: Window Operators sending directory instead of fully qualified file?

2016-07-16 Thread Clifford Resnick
Using Flink 1.1-SNAPSHOT, Hadoop-aws 2.6.4 The error I’m getting is : 11:05:44,425 ERROR org.apache.flink.streaming.runtime.tasks.StreamTask - Caught exception while materializing asynchronous checkpoints. com.amazonaws.AmazonClientException: Unable to calculate MD5 hash: /var/folders