Data retained after deletion of a ScaleIO volume
Certain storage volume configurations allow newly created volumes to contain previous data. This could lead to leakage of sensitive information between tenants.
Affected Services / Software
Cinder releases up to and including Queens with ScaleIO volumes using thin volumes and zero padding.
Using both thin volumes and zero padding does not ensure data contained in a volume is actually deleted. The default volume provisioning rule is set to thick so most installations are likely not affected. Operators can check their configuration in `cinder.conf` or check for zero padding with this command `scli --query_all`.
Operators can use the following two workarounds, until the release of Rocky (planned 30th August 2018) which resolves the issue.
1. Swap to thin volumes
2. Ensure ScaleIO storage pools use zero-padding with:
scli --modify_zero_padding_policy (((--protection_domain_id <ID> | --protection_domain_name <NAME>) --storage_pool_name <NAME>) | --storage_pool_id <ID>) (--enable_zero_padding | --disable_zero_padding)`
Contacts / References
Author: Nick Tait
This OSSN : https://wiki.openstack.org/wiki/OSSN/OSSN-0084
Original LaunchPad Bug : https://bugs.launchpad.net/ossn/+bug/1699573
Mailing List : [Security] tag on email@example.com
OpenStack Security Project : https://launchpad.net/~openstack-ossg