almost 80 fixes and improvements are done now and include:
NUTCH-2375 upgrade to new mapreduce API
It was a huge change affecting more than 10,000 lines of code. Thanks, Omkar!
Well, there have been some regressions but those are resolved now. Tests in
pseudo-distributed mode  succeeded and also a mid-size test crawl (180
million pages) on a Hadoop cluster.
Would be great if anybody is able to test the Nutch master in combination with
a non-HDFS file system (e.g. s3://)! Please let us know whether this works. Thanks!
NUTCH-1480: Multiple index writer instances with different configurations
Thanks to Roannel it's now possible to index into multiple Solr or Elasticsearch
instances. With NUTCH- (needs to be reviewed) also the routing to of documents
to the index will be configurable.
NUTCH-2583: Ralf contributed a huge upgrade of dependencies.
Nutch now runs and compiles on Java 9 + 10. Only errors in unit tests need
to be addressed in NUTCH-2596.
And two important issues are almost ready to be committed soon:
NUTCH-2549: a long list of fixes and improvements to protocol-http. Thanks to
NUTCH-2576: plugin protocol-okhttp, a new HTTP protocol implementation based
on the okhttp library. Supports HTTP/2.
The full list of fixes and improvements is available at .
I'll plan to work through the remaining 70 open issues during the next
days and hope to commit/resolve 15-25 of them and move the remaining
ones to Nutch 1.16.
Please vote for issues you want to get included. If there are open
pull requests, it will help if these can be merged, the unit tests
pass, and any review comments are addressed. Thanks!
If there are any objections or blockers, please also let us know!
I'll also plan to run a test crawl on Hadoop mid of this week.
But any help in testing is welcome.
Note that the tutorial needs to be updated (will be done after 1.15
is finally released) to reflect the changes related to NUTCH-1480.