H2O Import files via REST API from local server fails

Question

I am trying to use h2o REST API to import CSV files that I have on my local server.

Command:

curl -v -X GET 'http://127.0.0.1:54321/3/ImportFiles?path=http://127.0.0.1:8083/datasets/tables/csv/RDsTWgcvAjHeWJFnbhCKTCE5rn6aLCjJ.csv'

Result in following log:

Trying 127.0.0.1... Connected to 127.0.0.1 (127.0.0.1) port 54321 (#0) GET /3/ImportFiles?path=http://127.0.0.1:8083/datasets/tables/csv/RDsTWgcvAjHeWJFnbhCKTCE5rn6aLCjJ.csv HTTP/1.1 Host: 127.0.0.1:54321 User-Agent: curl/7.47.0 Accept: /
HTTP/1.1 200 OK X-h2o-build-project-version: 3.16.0.2 X-h2o-rest-api-version-max: 3 X-h2o-cluster-id: 1512722051559 X-h2o-cluster-good: true X-h2o-context-path: / Content-Type: application/json Content-Length: 349 Server: Jetty(8.y.z-SNAPSHOT)
Connection #0 to host 127.0.0.1 left intact {"__meta":{"schema_version":3,"schema_name":"ImportFilesV3","schema_type":"ImportFiles"},"_exclude_fields":"","path":"http://127.0.0.1:8083/datasets/tables/csv/RDsTWgcvAjHeWJFnbhCKTCE5rn6aLCjJ.csv","pattern":null,"files":[],"destination_frames":[],"fails":["http://127.0.0.1:8083/datasets/tables/csv/RDsTWgcvAjHeWJFnbhCKTCE5rn6aLCjJ.csv"],"dels":[]}

H2O log on TRACE level shows only:

12-08 15:41:59.951 10.8.128.101:54321 36013 #4756-331 INFO: GET /3/ImportFiles, parms: {path=http://127.0.0.1:8083/datasets/tables/csv/RDsTWgcvAjHeWJFnbhCKTCE5rn6aLCjJ.csv}

Is there any way to debug while importing fails? h2o does not asking local server at all.

Commands from other servers work well:

curl -v -X GET "http://127.0.0.1:54321/3/ImportFiles?path=http://s3.amazonaws.com/h2o-public-test-data/smalldata/flow_examples/arrhythmia.csv.gz" curl -v -X GET "https://raw.github.com/h2oai/h2o/master/smalldata/logreg/prostate.csv"

TomKraljevic TomKraljevic · Accepted Answer · 2017-12-08T13:45:53

In general, trying to interact directly with the H2O REST API isn't easy. The vast majority of people use a pre-made API client like Python or R.

But if you really want to do this, I would debug it by comparing with something that's working. Like the R client for H2O.

Write an R program that does this:

h2o.init()
h2o.startLogging()
h2o.importFile("/path/to/data.csv")

The startLogging() call will produce a detailed log file with all the REST API requests and responses. Look at that and try to mimic it.

You can also refer to the autogenerated REST API documentation (http://docs.h2o.ai/h2o/latest-stable/h2o-docs/rest-api-reference.html), but I would caution that if you tried to write a working client just based on the docs it would be hard.

Looking at a logged conversation from an already working client is by far your best bet.

H2O Import files via REST API from local server fails

3 Answers