Hello,
I am about to send 13 pull requests to hraven, on various changes that I have done:
Setting some context here before sending them.
I have been working adopting/using hraven at InMobi since a couple of months. While the storage of metrics to Hbase + the REST APIs have been useful for adhoc analysis of issues and having a more queryable job history than hadoop's, we needed a way to visualize / trend the metrics over time in graphite, and found hraven's robust handling of hadoop job history (both v1 & v2 support, identification of flows, app layer specific logic for pig and cascading, etc.) to be the perfect choice over other means.
Happy to announce that hraven has been running in production at InMobi for around 2 months, processing over 2 million hadoop job history files a month and sending over all the metrics to graphite. Over the course of hraven's adoption we have made some changes like refactoring it for multiple sink support, making it compatible to run on oozie/falcon, ability to have path exclusion/inclusion filters, hbase 0.90.x compatibility, bug fixes, etc.
Hope to get good cooperation from hraven-devs on accepting these changes to upstream.
Thanks,
Angad Singh
Senior Software Engineer, InMobi