ESX VSAN memory congestion

[Rand__]

Since we established in another thread that I don't have an alternative expect vsan for my soho use case I thought I'd dig into trying to debug the issue - or to uncover if I have an issue at all.

Surprisingly I actually found something which is not really looking good - unfortunately resources for interpreting this are quite sparse. Basically at one of my nodes there seems to be memory congestion which I can't really explain.



This box is the most loaded of my 3 but it should have sufficient resources...
Its an 2680v4/128 GB Ram, during the test it had 96GB Ram used.

Before I start a lengthy explanation of my setup and all that - is there a good resource where I can read up on this?


Edit: actually its not loaded at all at the moment...


I also shut down a freenas vm, that freed up 64gb but not effect on the congestion level... Don't think its Ram

 

[Rand__]

So I evacuated my vsan, removed and re-added all drives of all three hosts and now the problem is not happening any more (on empty vsan).


Edit: Too soon...
Ran another bunch of stress tests - new host, same symptom.
Looks like it happens on write benches, that points to the caching layer disk. Will replace the P320h with a spare P3700.