I would like to launch a debate or a discussion.
From my experience of 2 years and half now of this app, I don’t think our crashes problems are resulting from a third party program.
I think this is bad management in the source code of ressources management.
My crashes have always usually intervened when the RAM is eaten up.
I know it’s the OS responsability since the memory is all virtualized, to drop code segment from ram to swap etc. But there is one responsability which is delegated to the dev in C++ for example, it’s the allocation of memory. And I think there is a problem into the dynamic allocation of memory. Somewhere in the code, maybe the database code since it’s proprietary as declared earlier, there is a lack of memory at some point and so the value are changing from something to null because of corruption due to lack of memory.
Or the other possibility is that windows is that bad at managing the virtual memory that it doesn’t move code segment correctly, maybe due to an incompatibility with the database proprietary code.
Tell me what you think.
But from what I can see whether it’s my low end tablets device with 16G or any high end pc with 32G it’s always at the verge where ram is fully occupied that the database/core server is crashing are beginning to write corrupt data.
Even if some would consider that’s PC bad management, nowadays it shouldn’t matter how much process you have started because kernels are smart enough to manage. Only if you indeed need a low latency environment for some software like games, if it’s for développement or IT or whatever other works even virtual machines it should not matter. Plus in Linux set of mind, the idea has always been that RAM should at all time be fully occupied even if it’s by OS cache and clearly windows is doing that for a couple of years too now.