On Wed, 2009-06-10 at 18:18 +0000, Joseph L. Casale wrote:

I am rsyncing and remotely doing some work based on a logfile from a windows box from a centos backup server. I get the output from a vss snapshot that has a section like this:

• SNAPSHOT ID = {639ef5df-c933-4496-878a-ed57b9d52876} ...

  • Shadow copy Set: {427ac5db-21be-4c53-8ca4-24e7bac86a1d}
  • Original count of shadow copies = 2
  • Original Volume name: \?\Volume{787bdf7a-ccff-11dd-9866-806e6f6e6963}\ [D:]
  • Creation Time: 6/10/2009 12:12:08 PM
  • Shadow copy device name: \?\GLOBALROOT\Device\HarddiskVolumeShadowCopy11
  • Originating machine: milano.example.local
  • Service machine: milano.example.local
  • Not Exposed
  • Provider id: {b5946137-7b9f-4925-af80-51abd60b20d5}
  • Attributes: No_Auto_Release Persistent Differential

• SNAPSHOT ID = {fb5996cf-a35b-4110-b83c-45f9c64a75f3} ...

  • Shadow copy Set: {427ac5db-21be-4c53-8ca4-24e7bac86a1d}
  • Original count of shadow copies = 2
  • Original Volume name: \?\Volume{787bdf7b-ccff-11dd-9866-806e6f6e6963}\ [E:]
  • Creation Time: 6/10/2009 12:12:08 PM
  • Shadow copy device name: \?\GLOBALROOT\Device\HarddiskVolumeShadowCopy12
  • Originating machine: milano.example.local
  • Service machine: milano.example.local
  • Not Exposed
  • Provider id: {b5946137-7b9f-4925-af80-51abd60b20d5}
  • Attributes: No_Auto_Release Persistent Differential

My current script only uses sed/grep but I never had to deal with this case where now I need to do something based on the grep'ed/sed extraction of the value of "SNAPSHOT ID" based on volume name "D:" or "E:"? To make my script work without rewriting it, is there a sed method to only take the "SNAPSHOT ID" if the text indented beneath it has "D:", then do the same and extract "SNAPSHOT ID" if and only if "E:" follows?

Are the volume IDs in just two groups? E.g. all "C:", then all "D:", "E:', ...?

If so, use of the csplit command as a precursor to the sed should really shorten the learning curve and be more robust against any unexpected format changes.

If that doesn't look feasible or desirable, post a small snippt as an example and I'll see what I can do.

Others will suggest perl and (g)awk, etc. Both can be good solutions, depending. If the csplit wil work within your design, it would probably be easier due to the utility nature and very small learning curve. "Man csplit".

Thanks for any help! jlc

<snip sig stuff>

HTH