CODEWITHSUNDEEP
pysparkdatabricks
Sains
Sains

Reputation: 477

Databricks dbutils.fs.ls shows files. However, reading them throws an IO error

I am running a Spark Cluster and when I'm executing the below command on Databricks Notebook, it gives me the output:

dbutils.fs.ls("/mnt/test_file.json")

[FileInfo(path=u'dbfs:/mnt/test_file.json', name=u'test_file.json', size=1083L)]

However, when I'm trying to read that file, I'm getting the below mentioned error:

with open("mnt/test_file.json", 'r') as f:
  for line in f:
    print line

IOError: [Errno 2] No such file or directory: 'mnt/test_file.json'

What might be the issue here? Any help/support is greatly appreciated.

Upvotes: 8

Views: 50583

Answers (2)

AKS KRISH
AKS KRISH

Reputation: 1

you could directly use

dbutils.fs.head("file path")

if you intended to view the data.

Upvotes: 0

Silvio
Silvio

Reputation: 4207

In order to access files on a DBFS mount using local file APIs you need to prepend /dbfs to the path, so in your case it should be

with open('/dbfs/mnt/test_file.json', 'r') as f:
  for line in f:
    print(line)

See more details in the docs at https://docs.databricks.com/data/databricks-file-system.html#local-file-apis especially regarding limitations. With Databricks Runtime 5.5 and below there's a 2GB file limit. With 6.0+ there's no longer such a limit as the FUSE mount has been optimized to deal with larger file sizes.

Upvotes: 16

Related Questions