With the prior version (also called Beta 2) of VSS 8.0, an analyze using the usual V6.0 options (-f -c -d) took 5 hours. Now it takes over 12, and results in thousands more errors and bad files. There appear to be many new options available in this version. What do they do, and what affect will they have in "fixing" the database. Thanks

New options in Latest Beta 2(Build 8.0.50215.44) Release of VSS 8.0
elygirang
Thanks
Print
Thanks a TON, this changed the performance/behavior significantly for the better. These options/interactions need to be well documented by the fianl release though, otherwise many of us will continue to be stuck with a V6.0 methodology and mindset. PS#1, there may be a bug/conflict in the option processing since the way I was running ("-f -v3 -d -c -dc -fl -fp -i-") DID NOT delete the non-VSS files (XYZ.old) we had within the data\a..z\ folder tree. That was one of the reasons there were always so many errors; the *.old's were relisted every time(filed as bug). PS#2, now that I can see it, the SharedList.log file is a nice touch.
Thanks Again
bberndt
Baskar V
By default, analyze from VSS2005 uses a different reference checking method than VSS6 and may find more errors in the database; you can specify -refv6 to use the old algorithm.
You can run "analyze - " to see a summary of the accepted parameters.
However, none of them were added during latest builds of Beta2.
What is the build you're comparing with Is it possible that you were comparing with a CTP build
I would also recommend using the -c option less frequent. With it specified, VSS will rebuild to compact all the files in the database, so it will definitely require a long time to complete. It's just a personal option, but I'd rather live with some orphaned comments in the files taking a couple hundred bytes than wait a couple of hours for the full database rebuild.
Thanks,
Alin
Prudie
I think you're mistaken when assuming that every file in the backup folder is an "error". I suspect your database is already fine, but you're getting all those files in the backup folder just because you're specifying parameters you don't need.
For starters, using "-v3 -i-" may not be a good idea if you're interested in seeing what's wrong with your database. Using these flags suppresses many log entries.
I'd use -v4 instead.
As I told you before, passing -c will rebuild the whole database. Before rebuilding a file, the old one is copied into the backup folder (so no wonder you end up with 58k files there). I'd stop passing in this flag for now.
Yes, analyze.bad contains a list of filenames and actions. If you know what you're doing you can run "analyze -x" and later "analyze -f @analyze.bad" to detect and fix errors in 2 steps.
backup\Labels folder contains a backup copy of the data\labels folder before analyze was run. It was created because you specified the -fl option whick used the VSS6 method of rebuilding the labels.
VS2005 rebuilds by default the labels (unless you specify the -refv6 argument) during the new-style reference checking, and this is when the LabelCache folder is created. With the command line you're using, you spend twice the time to rebuild the label cache (once using the VSS6 style, once using the VSS2005 style).
You don't need to specify -fl if you're not specifying -refv6.
-fp flag is likely unnecessary, too. When this is passed in, analyze looks in the data\X folders for files that are not valid physical files (not of the format BCDAAAAA) and deletes them. Normally there are no such files, and even if they were, they should be already gone (if you already run analyze 8 times)
No, the case difference in analyze.log and analyze.bad has no significance.
To see if your database has any errors, I'd only pass the "-v4" flag to analyze.
Then, if there are errors, I'd use the "-f -v4 -d" flags.
Alin
Prasanna vignesh r
Thanks you for reporting the .old files problem. I've found this bug, too, 2 days ago when I played with analyze. The bug is not in the options processing, but in what actions does analyze take when those options are specified.
It seems that the .old files are regenerated everytime "-c" option is specified for analyze, and they are never cleaned, not even when "-d -fp" are passed alone.
I opened a bug for this and the current thinking is that it will be fixed in the final VSS 2005 version.
Thanks,
Alin