Reindex fails - socket hang up on initial search

[jimpaintertnf]

Node v5.7.1
Elasticsearch-reindex v1.1.14

Attempting to reindex from one index into another, using the following:

elasticsearch-reindex -f https://hostname.es.amazonaws.com/originalindex/log-type -t https://hostname.es.amazonaws.com/newindex/ index-script.js

Results in the following error:

Elasticsearch ERROR: 2016-08-11T15:37:35Z
Error: Request error, retrying
POST https://hostname.es.amazonaws.com/originalindex/NCSA-common-log-format/_search?search_type=scan&scroll=1m&size=100 => socket hang up
at Log.error (/temp/.nvm/versions/node/v5.7.1/lib/node_modules/elasticsearch-reindex/node_modules/elasticsearch/src/lib/log.js:225:56)
at checkRespForFailure (/temp/.nvm/versions/node/v5.7.1/lib/node_modules/elasticsearch-reindex/node_modules/elasticsearch/src/lib/transport.js:240:18)
at HttpConnector. (/temp/.nvm/versions/node/v5.7.1/lib/node_modules/elasticsearch-reindex/node_modules/elasticsearch/src/lib/connectors/http.js:162:7)
at ClientRequest.wrapper (/temp/.nvm/versions/node/v5.7.1/lib/node_modules/elasticsearch-reindex/node_modules/lodash/index.js:3095:19)
at emitOne (events.js:90:13)
at ClientRequest.emit (events.js:182:7)
at TLSSocket.socketCloseListener (_http_client.js:271:9)
at emitOne (events.js:95:20)
at TLSSocket.emit (events.js:182:7)
at TCP._onclose (net.js:475:12)

The same result if the last argument (the custom index script) is omitted.

Note the the URL being POSTed to in the log above is accessible via curl and returns

{
"_scroll_id": "somestuff==",
"took": 17,
"timed_out": false,
"_shards": {
"total": 5,
"successful": 5,
"failed": 0
},
"hits": {
"total": 176434,
"max_score": 0,
"hits": []
}
}

[vlaskinvlad]

+1

[danpaz]

I have experienced the same and I think it may be an underlying elasticsearch.js issue. Possibly it's related to elastic/elasticsearch-js#433. FWIW when I set --sniff_cluster false I don't see these request errors anymore, so you may want to give that a try.

[OriHoch]

I'm experiencing the same, --sniff_cluster false as suggested above by @danpaz worked for me as well 👍

[scotttam]

👍 I'm also experiencing the same problem and setting --sniff_cluster false worked for me too.