Support transient SMART failures

[dcelasun]

As discussed in #374 some SMART errors are transient and should not be treated as permanent.

This commit adds support for a configurable list of ATA SMART attribute IDs, failures of which will be treated as transient. Drive health history is still recorded and notifications are sent, but the device itself is not marked as failed.

Fixes #374.

--

@AnalogJ apologies for the cosmetic changes, they are made automatically by goimports. I can revert them if you'd like.

[AnalogJ]

Thanks so much for the PR 🥳

Apologies for taking so long to get to this, I've been a bit distracted by some other projects.
TBH, I think I would have implemented this slightly differently. Rather than "ignoring" the failure at the attribute processing stage, I think i would allow it to mark the attribute entry as "failed" however, I would filter out the failure before setting the DeviceStatus -- which persists across multiple SMART collector runs.

https://github.com/AnalogJ/scrutiny/blob/master/webapp/backend/pkg/web/handler/upload_device_metrics.go#L53-L59

Basically since the value fluctuates (and recovers) constantly, a failure at a single point in time should still be marked as a failure, but shouldn't cause the disk to the permanently marked as "bad"

Hopefully that makes sense?

[dcelasun]

Hopefully that makes sense?

It does. Let me make some changes :)

[dcelasun]

Apologies for the delay on this. Now that I look at it again:

Basically since the value fluctuates (and recovers) constantly, a failure at a single point in time should still be marked as a failure, but shouldn't cause the disk to the permanently marked as "bad"

Isn't this already the case? The Status field in the Smart` type refers to the device's status, not the attributes, and my PR simply doesn't update that field for transient failures.

[enoch85]

Would be great if you could have another look at this. 🙏