Google Cloud Storage: Mismatch In Crc32c & Md5 While Upload String To Gcs

While trying to upload the JSON string and overwrite the existing object in the GCS bucket, getting the below error. google.api_core.exceptions.BadRequest: 400 POST https://storage

Solution 1:

To replicate the error you have encountered:

import json
from google.cloud import storage

client = storage.Client()
bucket = client.get_bucket('some-bucket')

# blob1 object
blob1 = bucket.get_blob('file.json')

# downloads content
blob1_string = blob1.download_as_string()

# converts to dict and update content
blob1_obj = json.loads(blob1_string)
blob1_obj['some-key'] = 'some value'# upload using same blob instance
blob1.upload_from_string(json.dumps(blob1_obj))

# throws error like this `Provided MD5 hash "Ax9olGoqOSb7Nay2LNkCSQ==\" #doesn't match calculated MD5 hash \"XCMPR0o7NdgmI5zN1fMm6Q==\".",

You're probably using the same blob to download and upload contents. To prevent this error you need to create two instances of blob:

import json
from google.cloud import storage

client = storage.Client()
bucket = client.get_bucket("some-bucket")

# blob1 object -- for downloading contents
blob1 = bucket.get_blob('file.json')

blob1_string = blob1.download_as_string()
# Convert to dictionary
blob1_obj = json.loads(blob1_string)
# Add stuff
blob1_obj['some-key'] = 'some value'# blob2 object -- for uploading contents
blob2 = bucket.get_blob('file.json')

blob2.upload_from_string(json.dumps(blob1_obj))

# no error

Solution 2:

Just in case if anybody needs this 9 month later. Using of two different blobs is not you typically want. Very often you have to do many reads and writes in both directions. So I would strictly discourage using this approach. You need just to refresh the CRC32 checksum explicitly by calling "reload()":

storage_client = storage.Client()
bucket = storage_client.bucket(bucket_name)
config_blob = bucket.get_blob(destination_blob_name)
config_blob.reload()
config_blob.upload_from_string(json.dumps(config_data,indent=4), 
content_type='text/plain')