diff options
| author | Bryan Newbold <bnewbold@archive.org> | 2020-02-13 23:11:14 -0800 |
|---|---|---|
| committer | Bryan Newbold <bnewbold@archive.org> | 2020-02-13 23:11:14 -0800 |
| commit | f613f69a40fcc9a445f21cadd35d7c36c8061db8 (patch) | |
| tree | 30bbbc7a3463625adf9b567cb91623c917ea5562 | |
| parent | 2db64d580e3b5995e328c719e2bd014124179d0b (diff) | |
| download | sandcrawler-f613f69a40fcc9a445f21cadd35d7c36c8061db8.tar.gz sandcrawler-f613f69a40fcc9a445f21cadd35d7c36c8061db8.zip | |
ingest: bulk workers don't hit SPNv2
| -rwxr-xr-x | python/sandcrawler_worker.py | 2 |
1 files changed, 2 insertions, 0 deletions
diff --git a/python/sandcrawler_worker.py b/python/sandcrawler_worker.py index 02d075c..0ba4d03 100755 --- a/python/sandcrawler_worker.py +++ b/python/sandcrawler_worker.py @@ -108,6 +108,8 @@ def run_ingest_file(args): grobid_client=grobid_client, sink=sink, grobid_sink=grobid_sink, + # don't SPNv2 for --bulk backfill + try_spn2=not args.bulk, ) pusher = KafkaJsonPusher( worker=worker, |