Potential History file corruption?

[Alex Kiernan]

Roman Gollent <roman at pobox.com> writes:

> Hi folks,
> 
> We're in the process of replacing our old news router with a
> new machine running INN 2.3.2.  We're using CNFS exclusively
> with 3 large raw partitions (30gig, 20gig, 1gig) on a
> hardware raid array.  The machine in question is an E220
> with 512 meg of RAM and an A1000 array running Solaris 8
> 04/01.  We've encountered a problem where in the first hours
> of operation, grephistory returns the correct token for a
> given message id.  After a while however, grephistory ceases
> to work returning the first line of the history file:
> 
> 12:01:00 news at nostromo:~/db% grephistory -l '9iui8k$q07$1 at news.onet.pl'
> [EA854C9986078812E19714E107EDD2F4]	995240666~-~995080854	@030254574F0000000000000C19AC00000003@
> 
> compare this to:
> 
> 12:01:13 news at nostromo:~/db% head -1 history
> [EA854C9986078812E19714E107EDD2F4]      995240666~-~995080854   @030254574F0000000000000C19AC00000003@
> 
> furthermore, we actually know the correct token (having
> looked at the news log), and a grep of the text history file
> reveals:
> 
> 12:01:51 news at nostromo:~/db% grep 030254574F000000000000681C3200000003 history
> [34137FEA96920933DAFAE5858C7739D5]      995310036~-~995282070   @030254574F000000000000681C3200000003@
> 
> So the entry does exist.  This leads us to believe that at
> some point in time something goes horribly awry in the
> history db files.  Has anyone else encountered this problem?
> 

Do you still have the history* files? If you could mail me copies (or
put them somewhere I can grab them from) I'll try & take a look.

-- 
Alex Kiernan, Principal Engineer, Development, Thus PLC