image-copy should hash location and subscription for storage account name

[Deshke]

Extension name (the extension in question)

image-copy

Description of issue (in as much detail as possible)

https://github.com/Azure/azure-cli-extensions/blob/master/src/image-copy/azext_imagecopy/create_target.py#L32

• australiasoutheast + subscription_hash[:7] is longer as the limit for storage accounts

dirty work around in the create-target.py

    subscription_hash = hashlib.sha1(subscription_id.encode("UTF-8")).hexdigest()
    location_hash = hashlib.sha1(location.encode("UTF-8")).hexdigest()
    unique_storage_account_string = location_hash[:11] + subscription_hash[:11]

    # create the target storage account
    logger.warn(
        "%s - %s - Creating target storage account (can be slow sometimes)", location, unique_storage_account_string)
    target_storage_account_name = unique_storage_account_string

should be uniqe enough, while still under the limit of 24 chars

[tamirkamara]

@Deshke My original thought was to include the actual location name from tractability - I don't like all those totally random resource names. Probably didn't do the math right for all the regions, my bad.
How about keeping the actual location and adding a substring of the location subscription hash to complete to 24 chars?

[Deshke]

australiasoutheast is 19 chars, so you can shorten the subscription hash down to 5 but the problem may come up again if a subscription id is similar to one from another account, because all azure customers share the dns name.

also there is a artifical limit to the snapshot name

   # Create the snapshot in the target region from the copied blob
    logger.warn(
        "%s - Creating snapshot in target region from the copied blob", location)
    target_blob_path = target_blob_endpoint + \
        target_container_name + '/' + blob_name
    location_hash = hashlib.sha1(location.encode("UTF-8")).hexdigest()
    source_os_disk_snapshot_name_hash = hashlib.sha1(source_os_disk_snapshot_name.encode("UTF-8")).hexdigest()
    target_snapshot_name = source_os_disk_snapshot_name_hash[:11] + location_hash[:11]
    cli_cmd = prepare_cli_command(['snapshot', 'create',
                                   '--resource-group', transient_resource_group_name,
                                   '--name', target_snapshot_name,
                                   '--location', location,
                                   '--source', target_blob_path])

had to add the hash thing there as well or i could not create within eastasisa and australiasoutheast

[tamirkamara]

I might be missing something but subscription id is a guid. I have access to a bunch of subscriptions with all ids very different from one another. In theory I agree it can happen, but I don't think it's likely. Anyway, I can change that.

About the snapshot name - I'll limit it to 80 chars. No need for hashing since uniqueness isn't required here.

Anything else you saw?

[Deshke]

nope that should be it.

azure respons with invalid name if the snapshot name is to long

[Deshke]

eh @tamirkamara

About the snapshot name - I'll limit it to 80 chars. No need for hashing since uniqueness isn't required
here.

the snapshot name has to be uniq in the resource group, so cutting of after 80 chars may produces some other issues