how does history file get too large?

[Russ Allbery]

Anne Wilson <anne at unidata.ucar.edu> writes:

> If there was something that would just check the size of the file and
> warn me if it's getting too big, then I could deal with it before it
> becomes a problem.  news.daily?  expire?  Or, maybe that happened and I
> missed it?  Just seems like there should be a better way to deal with
> this problem besides the crash and burn approach.  (The only error
> message was "ME source lost . Exiting")

Yeah, I agree, we should have better error handling here.  I hope to
eventually have a better history file format too; there's really no reason
now for the history file to be textual, since it's actually a fixed-length
field database for the most part and it's not like the entries are
human-readable anyway.  That would reduce the size of the history file by
about half.

> If I was to write such a script, could I just delete expired entries
> from the history file, remake the indexes and overview, and have INN run
> successfully?

Yes, if the script does the following in this order:

  Throttle innd.
  Build a new history file (maybe as history.n).
  Run makedbz on the new history file.
  Replace the old history database with the new one.
  Unthrottle innd.

It's important for innd to be throttled during the whole period of time
while you're changing the history file.

-- 
Russ Allbery (rra at stanford.edu)             <http://www.eyrie.org/~eagle/>

    Please send questions to the list rather than mailing me directly.
     <http://www.eyrie.org/~eagle/faqs/questions.html> explains why.