I agree with your assessment and proposed algorithm outline.
However, as long as MAAS silently falls back to zero-filling - that will not be something we can use for our use-case:
Issue #1 : Silent fallback to zero filling does not fit our use case
Issue #2 : NVMe not supported properly.
This will solve Issue #2 which is good but Issue #1 will remain unresolved.
A good compromise would be if the NVMe erase script is implemented as MAAS builtin destructive storage testing script. In such case, I would expect from script to fail on error and not perform any automatic fallback to zero-filling.
Hello Guilherme,
I agree with your assessment and proposed algorithm outline.
However, as long as MAAS silently falls back to zero-filling - that will not be something we can use for our use-case:
Issue #1 : Silent fallback to zero filling does not fit our use case
Issue #2 : NVMe not supported properly.
This will solve Issue #2 which is good but Issue #1 will remain unresolved.
A good compromise would be if the NVMe erase script is implemented as MAAS builtin destructive storage testing script. In such case, I would expect from script to fail on error and not perform any automatic fallback to zero-filling.
Regards,
Igor