Subj : hpt squish corruption
To   : Phil Kimble
From : Matt Bedynek
Date : Wed Jun 13 2018 21:04:48

Phil,

Tuesday June 05 2018 12:32, you wrote to All:

 PK> I recall a few years ago experiencing a similiar problem with hpt &
 PK> squish msg bases.

At some point in the past I have ran hpt against squish bases without issue.  
But, I have always maintained a strict processing work flow which ensured that 
no two things which might alter the message base files would access them.  By 
alter, I mean pack or renumber.  Appending to is fine.

For instance, when I did pack message bases, the tosser would never be running 
and I made sure to never invoke more than one tosser instance.  Some possibly 
use the binkd exec feature and if that is used must at minimum use the hpt lock 
feature in the event two instances were spawned in quick succession...

These days I run JAM but as far as process workflow I go a step further and 
kick off processing only by cron job processing at minimum once a minute and 
not at all if no nothing is received.  Due to the abundance of space thanks in 
part to modern storage I don't bother pack the message bases at all.  Fidonet 
can likely grow to the end of its existence and never fill up the measly 100GB 
of space allocated to it (currently at 16%).

Take care,

Matt

---
 * Origin: The Byte Museum - www.bytemuseum.org (1:19/10)

.