We crashed last Friday, 02/13/04 (it would be on Friday the 13th!), and a subsequent 
series of `guide`s (wufa) indicated that I had 22 damaged UniData files. I was able to 
`fixfile` about half of them; the other half had group errors, such as:
 
"Group 65056, block 65057 has incorrect group number 0". 
 
When I try to run dumpgroup, I got:
1:blk check error in U_read_group for file 'STUDENT.ACAD.LEVELS', key '', number=65056
1:U_read_group error in dumpgroup for file 'STUDENT.ACAD.LEVELS', key '', number=1
 
When I made an attempt to clear the group, I got:
fixgroup STUDENT.ACAD.LEVELS 65056
Fixgroup STUDENT.ACAD.LEVELS 65056 will make group 65056 empty,    
        do you wish to do it ?      [y/n] y
No input file is specified.
Group 65056 will be cleared. Do you really want to proceed? (Y/N) y
1:blk check error in U_read_group for file 'STUDENT.ACAD.LEVELS', key '', number=65056
read error,errno=0.
 
Needless to say, the file remained unchanged.
The effect was that various LIST, SELECT, COUNT and FILE.STAT commands were 
terminated, as well as thier UniBasic equivilents, with similar errors.
All the damaged files were dynamic files.
We're running UniData 6.0.3 under Solaris 8.
 
I was able to recover, at least partially, by the incredibly kludgy and ugly (and time 
consuming) method of using dumpgroup to dump all the other groups, and then using the 
resulting file names as keys to build a savelist, which I then used to copy everything 
but the bad group. I then was able to construct a list of the missing @ID's by 
comparing the savedlist with one constucted from my backups. All the missing records 
hashed to the damaged group. Any new ones which may have caused the damage in the 
first place were lost.
 
1. I'm sorry, but this was the best I could come up with after many hours. Please tell 
me there's a better way.
2. Why did dumpgroup/fixgroup fail? Isn't this what they're designed to do?
3. Is there another way to extract the keys from a damaged UniData file?
4. Is there a white paper or available technical document which describes, in some 
detail, the structure and layout of UniData files, which would aid me in writing my 
own utilities?
 
Thanks for any help.
 
John Cassidy
Unix Sys Admin
DCCCD
 


--
u2-users mailing list
[EMAIL PROTECTED]
http://www.oliver.com/mailman/listinfo/u2-users

Reply via email to