Block level restore

[mntbighker]

Is there a FAQ somewhere that details in what modes of operation SD is capable of block level restores? I have read several threads suggesting sparse image use and it is my understanding that you can't block restore from a sparse. So if you are using sparse, what is the process for converting the image and can it be done with SD? I assume converting means making the image ASR ready.

I have never used SD before. We went from CCC to Netrestore for image deployment and construction. The idea of keeping a live target image that can be updated without rebuilding is apealing.

[dnanian]

The only time we use a block-level copy is when restoring from an ASR-compatible DMG. Since 10.3.9, though -- as you've likely seen with other tools -- image conversion from sparse images to DMGs is problematic, and often fails. On top of that, ASR scanning also fails on occasion.

You don't *have* to create an image in SuperDuper! to use it, though. For example, you can create a Read/Write DMG in Disk Utility and then point us to it as a volume. Then, you can scan it with ASR for block restore.

If you use our direct DMG creation, realize that we can fail due to errors "below us" in the image frameworks. The end result is ASR compatible... but is read-only, and thus can't be updated.

All of that make sense?

[mntbighker]

I have experienced the flaky behavior you describe. But is it not true that block level (ASR) is the only method really practical for fast image deployment? Restoring from a sparse is all slow file level restore.. yes?

What I am really looking for is a practical workflow for maintaining and deploying images as efficiently as possible.

[dnanian]

It is true, no question. ASR's basically the only fast way to do block copying.

Most of our users aren't trying to image a whole series of machines from a single copy, so the issues we've all run into with this haven't come to the fore. What I'd do, in your case, is create a read/write DMG and then use "asr imagescan" to get it ready for ASR. That'll avoid conversion, and should work a treat...