Increasing memory usage when trying to reindex an index

[centic9]

I have an index with 26mio rows which I try to reindex. However memory usage of the reindex-script grows constantly, it is already at 4g when reporting 19% reindexing done. Is this expected, it seems a lot of memory for a simple copy-operation which should be done in a streaming fashion, right?

Also there are no documents arriving in index-2 yet, would it only start writing data at th end?

$ elasticsearch-reindex -f http://localhost:9200/index-1/type -t http://localhost:9200/index-2/type -c 3 -b 10000 -q 1000 -s 10m -o 600000
Starting reindex in 1 shards.
 reindexing [======------------------------] 5016000/26000000(19%) 4028.4 16852.6s - 1/1 working

 1119 ec2-user  20   0 4833m 4.1g 9904 R 99.8 13.8  68:04.02 /usr/bin/node /usr/bin/elasticsearch-reindex
2

[garbin]

No, I think this is a bug. It seems the target index cannot be wrote, you can checkout ./reindex.log, It should tell you what's wrong with that.