Member since
11-22-2016
14
Posts
23
Kudos Received
2
Solutions
My Accepted Solutions
Title | Views | Posted |
---|---|---|
1465 | 03-03-2018 12:26 AM | |
1352 | 03-02-2018 11:34 PM |
03-12-2018
02:19 AM
@Constantin Stanca I just tried it and it worked!!! I guess the metadata was corrupted. That is something that I will investigate later, but I am back in business before my Monday morning crowd gets cranky. Thank you @Constantin Stanca and @Jay Kumar SenSharma for helping me on a Sunday!
... View more
03-12-2018
02:16 AM
@Jay Kumar SenSharma After more investigation, I realized that other views were broken too and they seemed to have some issues with cluster configuration. I noticed that one of the IP addressed of one of the nodes was wrong. I checked the hosts file and IP addressed seemed fine. I restarted the ambari agent for that node and Ambari reported the correct IP this time. That did not resolve the problem. Good advice on setting the logging level. I'll change that and continue the troubleshooting.
... View more
03-11-2018
09:38 PM
3 Kudos
Ambari Files View indicates that service hdfs check failed screen-shot-2018-03-11-at-53312-pm.png I can't bring Files View. I go to Dashboard/HDFS and run a service check and nothing wrong is found. All blocks are fine: no corrupt, no missing. HDFS is not in safe mode. I restarted all HDFS services, restarted Ambari, the problem does not go away. I can access HDFS using CLI and everything is fine. The issue is that I have a few users that use that View. Any suggestion?
... View more
Labels:
- Labels:
-
Apache Ambari
-
Apache Hadoop
03-03-2018
12:26 AM
2 Kudos
@Boris Demerov I guess we are both data scientists. Storm comes with Python and Ruby. The right place to start is src/storm.thrift. Since Storm topologies are
just Thrift structures, and Nimbus is a Thrift daemon, you can create
and submit topologies in any language. Here's a specification of the protocol: Multilang protocol. The thrift structure lets you define multilang components explicitly
as a program and a script, e.g., python and the file implementing your
bolt. Multilang uses json messages over stdin/stdout to communicate with the sub-process. Python supports emitting, anchoring, acking, and logging. Storm "shell" command makes constructing jar and uploading to nimbus easy. Here is a good reference: https://docs.microsoft.com/en-us/azure/hdinsight/storm/apache-storm-develop-python-topology
... View more
03-02-2018
11:34 PM
4 Kudos
@Boris Demerov TensorFlow on HDP will be supported this year. I understood from a conversation with an Hortonworks engineer that running a GPU application like TensorFlow on HDP requires YARN enhancements. They are likely to be included in HDP 3.0, scheduled to be launched at DataWorks conference held in San Jose, June 2018. If you plan to attend DataWorks in Berlin (April 16-19, 2018), the conference schedule includes 4 presentations and one 2-day workshop in Deep Learning. You may get more updates from that conference and Hortonworks folks attending it.
... View more
03-02-2018
05:29 PM
5 Kudos
We are in process to migrate our HDP clusters to Google Cloud. Spark is heavily used by our applications and data scientists. GCS will be also heavily used to fetch raw data and post processed data. One challenge we have is the lack of support for Google Connector and capabilities around GCS and running Spark jobs efficiently on GCS stored data. This is not part of the HDP bundle and there is no information on whether this will be bundled and supported in the next versions. Does Hortonworks have a plan to include it in the following versions, also to support?
... View more
Labels:
03-08-2017
05:39 PM
@jwoodward Thank you.
... View more
03-07-2017
06:18 PM
Someone accepted the response on my behalf. That is not fair. This should not happen. I am entitled to my question and accept the response. Please stop this practice. I'm trying to UNACCEPT and that action is not even available. I will report this to HCC moderation to address. Please allow a submitter to unaccept a response and also stop this practice to accept response on my behalf. I asked the question, I should be able to test the response and accept the response was helpful to my problem. In this case both responses were good, but it is my choice to decide when to accept it.
... View more
03-07-2017
06:17 PM
1 Kudo
@Deepak Sharma @Artem Ervits Duh! It was easy. Thank you. I voted-up both responses. I apologize for accepting just one answer based on the timestamp of the response. I believe Artem was a few seconds faster.
... View more
03-07-2017
05:26 PM
2 Kudos
Labels:
- Labels:
-
Apache Ambari
-
Apache Hive