10-04-2017 07:49 PM
I recall there is a local computation mode in Oryx 1.x
This is useful for us to play ALS with smaller data set and see how it performs without going into all the Spark, Kafka setup.
However; I cannot locate the releases (or git source). Where can I find it ?
Solved! Go to Solution.
10-04-2017 07:55 PM
That implementation is obsolete at this point, I'd say, but sure you're welcome to go dig it out. It worked well.
The releases and source are still on the 1.x project site:
10-04-2017 08:08 PM
Oh, I forgot, we have made many obsolete repos in github.com/cloudera private. I can still see it but of course you can't.
Here's a tarball of the final release:
If there's any significant demand, I could revive the repo in my personal account
10-04-2017 08:19 PM
Got it and downloaded it.
It would be nice if you can revive the repo. I used Oryx 1.0 before and it worked very well.
If you cannot make the repo in short term, can you pass along the latest release jar files ?
10-06-2017 12:05 AM
Are you looking for the .jar files that were produced as part of the release? those are still in the repo and will stay there indefinitely as far as I know, just because it could be part of people's builds:
10-24-2017 04:06 PM
Two more questions about Oryx 1:
It has "model.local-data=false"
I set it to false to use HDFS
I am thinking "model.instance-dir=..." will point to HDFS.
My question is whether it supports S3? Something like "model.instance-dir=s3://bucket/..."
(2) Is it running with Java 7 as well (in local computation mode) ?
10-25-2017 02:52 AM
Even I have forgotten exactly how it works off the top of my head, but yes, you are correct that you should be able to use HDFS paths. Yes it runs on Java 7 -- or 8, I believe, though I don't recall if that was tested. It doesn't require Java 8.
10-25-2017 10:16 AM
Not clear what you mean "...you should be able to use HDFS paths....". Yeah, I know that for sure HDFS path. However, I am refering to S3 path, somthing like "s3://bucket/....".
Is S3 path possible ?