While wondering why my FM solution routinely goes down every few days, I copied it over to my test FM15 (Mac OS Server). I can semi-reliable cause my solution to crash the server by using the record selector on a layout which is linked to a 600 record shadow table. I randomly select a few various records and then I get the dreaded SDOD - "Sort dialog of death".
When I eventually get the redundant SDODs I watch the TopCallStats.log (see attached). While my FM Pro client is still running I get infinite ODBC Query lines. When I have to force quit FM Pro client The ODBC Query lines still continue but it replaces my name with an id number. This goes on and on without anyone connected to the server.
At this point I can use Admin Console to close the file. But, the ODBC Queries are happening still! There is no hosted file anymore. I have to restart FM Server to get them to stop.
On my FM 14 production system, I normally *can't* cleanly close the file because I suspect too many hung processes are going on from all the different users. If just *one* person gets the SDOD, then everyone will be hung up.
In my attempts to debug this, I'm making the layout more simple and removing any triggers. Another possibility is that a shadow table record is corrupt (but all data looks clean) and can somehow causes the client to go SDOD which in turn can crashes the server. Fun stuff....
Message was edited by: Eric Wood