At 07:13 PM 12/18/04, you wrote:
Has anyone else run into issues when converting UniData index files with convidx?
I'm preparing to assist a client with a switch from Tru64 to Solaris which use different byte ordering. The client uses RFS and has a number of large dynamic files and makes significant use of UniData indexes.
We've found that apart from the time involved in running convdata against the data files there aren't any problems with the data, but that convidx runs much slower than convdata, sometimes doesn't complete at all, and that some of the indexes appear to be corrupt after convidx has been run.
One large dynamic (recoverable) file had several idx00n partitions, and convidx appeared to hang (over three hours cpu time consumed) after processing the first idx001 file and had to be killed and the index completely rebuilt.
Other index conversions appear to have gone through OK, but then we've seen application issues when using the index. guide_ndx doesn't always seem to identify the problem, occasionally mentioning a key that is only in the index, but still we've seen UniData actually crash because a tm process dumped core while searching one of these indexes.
Are there known problems with convidx and is it best not to migrate the indexes, but instead rebuild them, or are there things I need to do differently in my migration process?
Best Regards,
Ken
PS. Xmas Greetings to all ------- u2-users mailing list [EMAIL PROTECTED] To unsubscribe please visit http://listserver.u2ug.org/
Kathy Tymoczko
Computing Services
Wabash College, Crawfordsville, Indiana 47933
[EMAIL PROTECTED] 765-361-6440 -------
u2-users mailing list
[EMAIL PROTECTED]
To unsubscribe please visit http://listserver.u2ug.org/
