System disk errors when running "analyze/disk"

[Hoff]

This disk is corrupt. Based on your "some issues" comment, it would appear that whomever installed this might have made a configuration mistake with clustering, and then either didn't notice the error or didn't correct it. If the system is booted with incorrect votes or expected votes values, this is the sort of corruption that can arise. And it can hit any disk that was mounted at the time.

The central question here is "how corrupt is this disk?" and that's not something we can answer. The same thing that happens to the disk blocks within the files of the ODS volume data structures can happen within other files elsewhere on this disk; what are effectively corrupted blocks can be spread across the spindles here.

The alternative is to start diff'ing files -- effectively all of them -- against the OpenVMS distro kit and the patch kits and the layered product distributions and known-good application copies. Not an easy task.

Or you can operate until something somewhere hits a bad block and blows up, and then recover that file and continue.

Here, I'd typically tend to reinstall, and reconstitute. Recovering from known-good BACKUP, and recovering from distribution kits. I might well ask for the original folks to come back and look at this particular corruption.

[Willem Grooters]

@rob:

The structural information of an indexed file is embedded in the file itself - and, observing the file as an object, it can be considered "content". ANALYZE/RMS will check the validity of the structure. This is what I intended to state.

I agree you can only check the structure, not the real data that matters. I don't think any tool will be able to without understanding what to expect. And there will be files that can never be checked on content (how would you tell an execuatble is corrupted by scanning it's content?)