At least it is know problem that splash "SQL has gone away" may appear after long uptime of Cqrlog. That is also more or less random. Many fixes are tried during years but without any breakthrough.

I am now struggling at the moment with strange SQL problem where I make a query giving a name the result of it. On return I check that given name exist in transaction before reading it, and when I am reading it I get error that the given name does not exist.
And to make it perfect it is randomly happening.

And RBNmonitor that does not loose it's connection to telnet.receresebeacon.net and gets spots, but looses it's internal connection to update them to RBNwindow's grid. And also this is random.

Somehow, sometimes I have feeling that FPC/Lazarus does not always work correctly. But that is hard to proof.

Quick fix: Save energy and shut of your computer while not being in shack... :-) That is all I can suggest for now!

If you ever get bullet proof way to reproduce any (non random) fault you find, please do not hesitate to report it.
I would be glad to fix it, if I can.