No, don’t worry, I think it’s totally compliant with Hadoop’s behavior but I wanted it to behave more like Flink (to totally clean the destination folder before outputing new files).
From: Fabian Hueske [mailto:email@example.com]
Sent: mardi 20 décembre 2016 16:41
Subject: Re: Generate _SUCCESS (map-reduce style) when folder has been written
Great to hear!
Do you mean that the behavior of Flink's HadoopOutputFormat is not consistent with Hadoop's behavior?
If that's the case, could you open a JIRA ticket to report this and maybe also contribute your changes back?
Thanks a lot,
2016-12-20 16:37 GMT+01:00 Gwenhael Pasquiers <firstname.lastname@example.org>:
Thanks, it is working properly now.
NB : Had to delete the folder by code because Hadoop’s OuputFormats will only overwrite file by file, not the whole folder.
The _SUCCESS files were originally generated by Hadoop for successful jobs. AFAIK, Spark leverages Hadoop's Input and OutputFormats and seems to have followed this approach as well to be compatible.
You could use Flink's HadoopOutputFormat which is a wrapper for Hadoop OutputFormats (both mapred and mapreduce APIs).
The wrapper does also produce the _SUCCESS files. In fact, you might be able to use exactly the same OutputFormat as your Spark job.
2016-12-20 14:00 GMT+01:00 Gwenhael Pasquiers <email@example.com>:
Sorry if it’s already been asked but is there an embedded way for flink to generate a _SUCCESS file in the folders it’s been writing into (using the write method with OutputFormat) ?
We are replacing a spark job that was generating those files (and further operations rely on it).