The customer was faced with some issue with sda and sdb being swapped during deployment.

it seems that it's the same issue:

https://bugzilla.redhat.com/show_bug.cgi?id=2252941

Fortunately, thanks to the well-written KCS(https://access.redhat.com/articles/7048606),

We thought the issue was resolved if it was created a symbolic link.

We have discovered some issue.

In swift ring, instead of the default volume being changed to swift0, it is recognized as swift0 being added to the existing volume, resulting in it is writing to both sides of the disk.

The swift uses both the existing mount area (/srv/node/sdb) and the new area (/srv/node/swift0), so a solution for this is needed.