[dale-cooper]

Maybe I'm missing something, but two thoughts:

1. Doesn't the overcommit feature lessen the benefits of this? Your initial allocation works but you can still run out of memory at runtime.

2. For a KV store, you'd still be at risk of application level use-after-free bugs since you need to keep track of what of your statically allocated memory is in use or not?

[nickmonad]

Author here! Overcommit is definitely a thing to watch out for. I believe TigerBeetle calls this out in their documentation. I think you'd have to explicitly disable it on Linux.

For the second question, yes, we have to keep track of what's in use. The keys and values are allocated via a memory pool that uses a free-list to keep track of what's available. When a request to add a key/value pair comes in, we first check if we have space (i.e. available buffers) in both the key pool and value pool. Once those are marked as "reserved", the free-list kind of forgets about them until the buffer is released back into the pool. Hopefully that helps!

[justincormack]

You can work around overcommit by writing a byte to every allocated page at allocation time, so that it has to be actually allocated.

[dnautics]

out of curiosity, does that generally mean that (linux) OOM killer can't get you? IIRC the oom killer is only triggered on new page request, and only the requesting process is eligble for the murder?

[chris-]

No, it does not. The oom killer acts on (mostly) the oom score and no process is exepmt, regardless of whether or not it allocates new memory. It may help you write correct programs in certain situations though, eg. if your program was running in a defined context, eg. a cgroup, and you would not allocate beyond your cgroup limits, and the system was configured sanely, you can handle allocation problems easier.

[dnautics]

aw damn. thanks for clarifying.