Reputation: 217
I'm unable to overwrite or remove a particular object from a Google Cloud Storage bucket that I own.
gsutil rm
fails with BadRequestException: 400 Invalid argument
:
% gsutil stat gs://mybucket/0e133e23-6005-4711-ac2e-24b2cc95ee05.jpg
gs://mybucket/0e133e23-6005-4711-ac2e-24b2cc95ee05.jpg:
Creation time: Fri, 07 Aug 2015 02:03:47 GMT
Update time: Fri, 17 Feb 2017 04:34:43 GMT
Storage class: STANDARD
Content-Length: 46169
Content-Type: image/jpeg
Metadata:
goog-reserved-file-mtime:1438913027
Hash (crc32c): UBUUbQ==
Hash (md5): fhU9YXjE7M6XRiT+05seeQ==
ETag: CPiV3t7wlccCEAI=
Generation: 1438913027083000
Metageneration: 2
% gsutil rm -a gs://mybucket/0e133e23-6005-4711-ac2e-24b2cc95ee05.jpg
Removing gs://mybucket/0e133e23-6005-4711-ac2e-24b2cc95ee05.jpg#1438913027083000...
BadRequestException: 400 Invalid argument
The GCP Console Storage Browser fails with Invalid argument (1 error)
and (digging deeper) storage_class_update_time_nano must be positive
.
The bucket is using the Standard
storage class with object-level ACLs enabled (although I'm not using them, and my account is listed as an owner for the object in question). The bucket does not have a retention policy, and the default event-based hold option is disabled. It does not have any lifecycle rules.
I can remove other objects in the bucket. I'm suspecting that this is a bug on Google's side, although I haven't been able to find reports online of others experiencing the same issue (e.g. Google returns zero results for the storage_class_update_time_nano
error message).
Upvotes: 1
Views: 923
Reputation: 217
This is resolved now: a gsutil mv
command that failed in the morning of 2021-02-08 worked when I ran it again four hours later.
In the GCP issue that I filed, a Google employee confirmed that it was caused by a regression:
I can confirm that the issue was on the GCP side and related to a new change on GCS which was performed in the beggining of February.
Now I can see that the issue was rolled out and fixed.
Upvotes: 1