INN not keeping up with incoming news

[bill davidsen]

In article <20010221235935.A4331 at opentransit.net>,
Fabien Tassin  <fta at sofaraway.org> wrote:
| 
| According to Alex Kiernan:
| > 
| > We've just taken to locking the history.{hash,index} in core; its
| > worked wonders
| 
| without touching to the INN code ?

  Don't need to touch the INN code.

| I haven't tried your program yet but I wonder if this job should not be
| done in dbz.c automatically. I also wonder about its behaviour during and after
| expire/shutdown/crashes/sync etc. If the files are never committed on disks,
| the risk is very high to corrupt or lose everything.

  Inndstart exists to avoid running innd as root, I don't see enough
benefit from building this into dbz.c to justify the security risks. The
posted program looks *very* much like the one I wrote...

  Depending on the o/s, a mmap()ed file will get written to backing
store, just not to swap. If you mlock() just some pages, that will not
get written anywhere. I don't have any more toy servers to crash one
deliberately and see what happens, but I suppose someone will get the
data for us eventually.
-- 
bill davidsen <davidsen at tmr.com>
  CTO, TMR Associates, Inc
Doing interesting things with little computers since 1979.