>Is there any kind of message being returned to the application from the
>crash? or are you saying that the application crash crashes the db server
>as well? i.e. that the Guardian restarts happen as a consequence of the app
>crashing the server?

Embarrassingly, the application is dying with messages. I obviously failed to
protect something important with a try-except. The crash is consistant with
connection failing though. As to whether this coincides with restart messages -
I hopefully will know more when ship contacts be again.

>Is this a single-box architecture? local connect, etc. Any devices
>attached? e.g. instrument interfaces, etc.?

Yes and yes! Single box and with a interfaces into the ship electronics. These
come in as NMEA packets into a serial port.

>Does the app guard against long-running transactions?

Hmm, not sure what you mean. try..finally blocks ensure transactions are
closed one way or another. The structure of the app ensures selects wouldnt
select more than 500,000 records in one transaction which don't take that
long.

>Could be - has anyone tried a backup + restore to a new file (not
>overwriting the database) to see whether there are some damaged structures?

No. Overnight the ship was replacing the current GDB with a backup "blank"
gdb to see what happens. If that worked, then I would try backup/restore but I
havent heard back.

>These are all shots in the dark. There don't seem to be many clues so far...

Yes. trying to fix a problem over Immersat is no fun. Setup is on quite a few
vessels but never had this problem before.

Thanks for your interest.


----------------------------------------------------------
Phil Scadden, Institute of Geological and Nuclear Sciences
41 Bell Rd South, PO Box 30368, Lower Hutt, New Zealand
Ph +64 4 5704821, fax +64 4 5704603