>How do you know that PowerMail is idle (i.e., not writing to the
>database) at the time your system is going down?
Because there is no activity. It is not tasked with receiving email or
sending it out, and therefore shouldn't be doing anything with the index
files. Also as you might know, the latest version appears to have turned
off background indexing (which is irrelevant anyway because we're talking
about hours unattended passing before my system crashes).
>Maybe there is an electrical problem with a hard drive, such that PM
>begins to check mail, spins up disk, starts to write, loses power.
That is grasping at straws.
>Of course I am purely speculating. But if a file is left in an
>inconsistent state on disk, it will result in data loss or at least
>necessitate maintenance; since that is the symptom, we can induce that
>for SOME REASON the database is in use or otherwise inconsistent when
>your system dies.
Correct. And the most likely reason is that something isn't being "put
to bed" correctly by PM. Something is being left untidy and when the
system goes down it stays that way. And why this should require a
complete, ground up rebuild (instead of a minor repair) also is a
question since Jerome seemed to imply that PM is capable of partial repairs.
>To give confidence in PowerMail's robustness, I just saved this message
>as I was typing it, then force-quit PowerMail point blank. I then
>restarted it, and up it came -- no database recovery or re-indexing required.
This doesn't mean anything. You would have to be using the same exact
system as me for this to be relevant. I can put a gun to my head and
pull the trigger with nothing happening. But if my gun is unloaded and
yours is... your results would be a lot different :-)
Note that the reason why I started participating on this list was due to
frequent 16 hour rebuilds. I found out that two interactive bugs with
the MacOS were responsible for these crashes and that the lengthy rebuild
is also unnecessary (in theory at least). Did you have these problems?
No, but I certainly did and Jermone's helpful tips fixed my problems.
>Therefore, absent your observation of what the system is doing when it
>dies, I would tend to believe that disk activity may well be going on.
There shouldn't be any activity, therefore there shouldn't be any
corruption. And even if there is activity, why is the entire 100,000
email Search index zapped into a useless heap of digital trash? That
doesn't sound right to me either.
Steve
