If a customer wishes to delete data, but there is not a deletion issue, i.e. it is a customer project to delete a large amount of data.
Please engage the PS team for that scenario.
A local team (i.e. accounts) can help with engaging the PS team.
Note that ECS support is for break-fix issues,
i.e. for bucket deletion issues and would need to prove to CS management of the issue.
Bucket deletion Issue investigation procedure:
- ECS Engineering would ask customers to attempt to delete the bucket themselves when possible, CS/OE intervenes to assist with bucket deletion because all previous attempts have failed with errors.
- An investigation would have to commence first by CS, due to the size of the target buckets/objects, please note, the investigation object scans may take time to complete.
- If there is dangling LS issue found on the bucket, CS/OE would require customer authorization to progress, CS would ask customers to explicitly state in an email that they want to delete the bucket/objects in a bucket.
- It must be clear for OE records that the customer want support to assist with data deletion issue. This would be reviewed by a ECS support colleague.
Example of a short statement in an email to CS during the SR investigation:
"We are certain we want to delete all of the objects in the bucket X in namespace Y
We would authorize ECS support to assist with requested deletion.
We had attempted to delete via our applications i.e. Geodrive, S3 browser, etc.
All attempts have failed. Please see the attached errors messages"
Please provide ECS support with the application errors messages for the SR investigation.
Please provide ECS support with a list of objects, if not all of the bucket is being deleted.
ECS support would require the bucket user ID and secret key to run several bucket investigation commands.
Please provide them when possible, or arrange a zoom session if you are unable to provide them via email.
ECS Engineering would want clear and explicit deletion authorization from the customer.
ECS Engineering would want to investigate the issue first, so please provide all possible information on the deletion issue.
After the Dangling LS scan is completed, if there is a dangling LS issue, CS would then proceed in clearing the Dangling LS, after which CS would ask for the customer to attempt the deletion again. It is always better that the customer delete data themselves.
If Dangling LS clearing is not the issue or if the customer was still not able to delete, then CS would look into deleting the object/bucket.
A
customer senior management would now be
required to give authorization to CS for CS/OE to delete data, instead of the customer deleting the data.
As CS/OE would be asked to delete customer data,
ECS support management would need to be informed and will confirm with the customer.
Summary:
1) Customer attempts deletion, whereby all attempts have not worked
2) SR opened and CS investigates, including a Dangling LS scan,
scan time is proportional to bucket size.
3) CS will ask the customer to email exact approval to clear possible dangling LS on the buckets if they exist
4) Customer to re-attempt deletion
5) If a customer is still unable, CS
will ask for deletion approval from customer senior management, and
CS will contact CS management for approval.
6) CS will attempt the delete on the
customers behalf, if the attempt is unsuccessful, CS will then engage OE to investigate.
Dangling LS may
more likely occur on systems with a high load with an OS version prior to 3.3.
Typically deletes in systems with high loads often resulted in OB entry being removed and an LS entry remaining, hence the term dangling.
3.3 and above has improved LS/MI migration performance that can help
mitigate dangling LS.
The dangling LS fix should not be impactful, but
please note may take time to resolve dangling LS on the bucket.