HCIBench stands for "Hyper-converged Infrastructure Benchmark". It's essentially an automation wrapper around the popular and proven VDbench open source benchmark tool that makes it easier to automate testing across a HCI cluster.
do you need vSAN Network diagnostic mode enable to have graphana gather mettrics ?
no, not needed, vsan performance service is good enough.
I cant find vmware command to clear read/write cache before the test. All my ESX host have different password and I want to clear cache manually
I would suggest standardizing the password on your esx hosts to insure your testing is fully automated and consistent. If you forget or incorrectly clear the cache between test instance you will compromise the integrity of your overall testing.
If you must do it manually, you need to manually copy the following tool from hcibench to each esxi host, then run it with the appropriate flags (--rd --wr)
What is the time of the test job when you use the Easy run option ?
How many hours or minute does it run ?
easy-run time = deployment(5mins) + vmdk_preparation(30mins to 2hrs) + N* single_test_case(30mins warmup + 60mins)
N is [1,4] depends on how many cases you selected when enabling easy-run.
And the easy run uses how many threads per disk ?
4k and 8k, 4threads/vmdk.
256k, 1thread/vmdk if checksum is on, otherwise 4threads/vmdk.
So the test itself is 60 mins per test_case ?
90mins in total, 30mins warmup + 60mins testing
I am testing agains a vSAN Datastore.
Why in Grafana HCIBench FIO Monitoring shows data whilst vSAN Overview doesn't (N/A) even once test is over?
I can see vSAN stats in vCenter but not Grafana.
how long was the testing running for? what vCenter version you are using?
It was an EasyRun running for quite a while (maybe 8 hours) as I left all 4 tests selected.
Seeing that I got the stats for cluster the second cluster I tests so I will try to test the first cluster again and check.
Is Congestion (apart from very high latency) a good indicator to see if I am overloading the vSAN cluster? Any other indicators to watch for?
What kind of congestion you saw? Would you be able to get into vSAN observer stats page and click the congestion on the upper right drop down list and get the details?
also, it matters what kind of data services you turned on for vSAN, like dedupelication, erasure coding etc.