What size recomendation for head node?

We have a license for MDCS 256 workers.

I've built the system, but the admincenter in particular seems to run quite slowly.

With 16 VMs with 16 vCPUs and 64GB memory each, what size should the head node be?

The head node I currently have is 4 vCPUs with 16GB and it feels sluggish.

Answer 1

[Dasharath Gulvady]

Does the head node support a minimum of 16,384 file file descriptors as mentioned in the documentation page of MDCS:

http://www.mathworks.com/products/distriben/requirements.html#mathworks-job-manager

As long as your head node meets the system requirements above, it should work fine.

Also, elaborate on what you meant by "it feels sluggish"?

[Lachlan Simpson]

Test Connectivity can take from 15-30 mins to complete, often fails. Doesn't log why it fails.

ulimit reports unlimited for bot users, but just to be safe, I've bumped nproc and nofile to 65k in /etc/security/limit.conf

[Lachlan Simpson]

! This answer not only worked, it solved another problem I had too.

Important to note that in Ubuntu, ulimit alone might say "unlimited" yet you see these results.

I used a combination of this post wrt hard coding limits and this post to discover that ulimit is quite rich and the man page is quite poor. (hint: ulimit -a tells everything)