VMWare network and vmxnet3 driver

Phenno
Phenno Member Posts: 630
in NAV Three Tier
Has anyone experienced issues with NAV on vmware virtual machine using vmxnet3 driver?

I'm currently investigating strange issues with NAS service on NAV2015 instance. I was just watching scheduled job doing delete of invoiced sales order (a lot of them to delete). This has been done out of working hours so both NAV and SQL servers aren't doing much in that period. While looking at processor time I'm seeing that NAV is doing some job for a minute, than it slowes down for several minutes. Than it goes up again, doing some work for a minute, than it slows down again for several minutes.

In period of doing some work it generates lot of network transfers (up to 15Mbps), while when he's doing slowly it generates no more than 355Kbps.

Also, I was checking progress via SQL and can see that in good period it deletes 20 or so sales headers per second, while in slow period it deletes very few.

Since I excluded cpu, memory, disk resource issues on NAV and SQL I'm looking at possible network issues with vmxnet3 network adapter. Once I read artikl mentioning issues with this driver so If anyone had similar experience, please share with us.

Article was here: https://kb.vmware.com/selfservice/microsites/search.do?language=en_US&cmd=displayKC&externalId=2008925

Also, I read similar article mentioning flooding SQL with NAS jobs, here (http://www.dynamics.is/?p=2487), though it applies to NAV2016.

Best Answer

Answers

  • Kishorm
    Kishorm Member Posts: 921
    Have a look at this blog post, Stryk says that it occurs with 2015 & 2016 so it maybe the problem you have...

    blog.stryk.info/2016/12/06/navsql-is-completely-stalled-because-of-too-many-cpu-threads/
  • Phenno
    Phenno Member Posts: 630
    edited 2016-12-16
    Yes, I already saw this and simptoms are similar but not the same. I did check workers count agains max workers count and numbers are not close. My max workers count is 544 (6cpu) and working threads are 80, mostly. Also, I do not see complete stall but rather slowdown that cannot be corelated to cpu, memory and/or disks.

    Btw. I saw that my instance tops 40 open connections (from NAV to SQL) although there is no such limit set anywhere. Is this fixed limit on NAV instance?

Categories