That’s my thinking too. And it might be impossible to track down.
1 Like
Not sure how it can be tracked and fixed. But it is a very serious problem needing serious research from Xojo, as it can potentially be causing data corruption and financial losses.
2 Likes
Could you count each query. Mesure
The time this middle ware is running until it break.
Can you setup a second middle ware just for testing with a test database and test app based on your original project.
About run some hour until it crash
Remember me at a wrong or buggy c++ runtime version but you said it happens since years.
To find the best reason try the method of elemination. This way you find it.
In past i had a rule … If me not found a bug in three days then it is not my fault.
The problem is, this might not manifest for weeks, or even months.
Maybe due to your defensive app reboot when idle for a long time (nightly maybe). But under a stress test it manifests faster.
Also can be tied to lots of thread creations and destructions with them “inside”. Something related to some accumulative condition like a resource leak.
1 Like
Could it be that the log get’s truncated (for whatsoever reason) in parallel(or even through) the triggering issue, perhaps pointing you to the wrong direction? Unlikely, but who knows …