[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Debian 9.2, VSphere 6.5.0, "TCP performance may be compromised"



I wrote:
> Right. But it's a big company and I can't influence when or how ESX gets upgraded.

Perhaps I should explain that everywhere I've worked that uses VMware, the VMware
admins are almost always current or former Windows admins. Naturally that's because
there are hundreds-to-thousands of desktops being driven out of VMware, and many now
are Horizon so they don't even have full, real hardware, so they're VMware-dependent.

As a result of these "facts on the ground", linux VMs do not have the "administrator mindshare"
that Windows VMs have with the VMware system admins (current/former Windows admins).
Not even if they are fundamental to the business.

As I mentioned, it's more sociology than computer science.

On Thu, Oct 19, 2017 at 1:51 PM, Nicholas Geovanis <nickgeovanis@gmail.com> wrote:


On Thu, Oct 19, 2017 at 1:25 PM, Sven Hartge <sven@svenhartge.de> wrote:
Nicholas Geovanis <nickgeovanis@gmail.com> wrote:

> And unfortunately the HTML5 version of the VSphere app doesn't permit
> editing of, for example, that vmxnet3 parameter for mitigation of
> https://bugzilla.kernel.org/show_bug.cgi?id=191201#c49

You don't need all that stuff if you are not on ESX 6.5. So no need to
try to edit those values. And if you upgrade to 6.5 eventually, just use at
least 6.5u1, then you are safe.

Right. But it's a big company and I can't influence when or how ESX gets upgraded.
I can only prepare my VMs for their new lives in this hard, cold, virtual world; and this
was the first attempt with Debian 9.2 on the current VMware release. Results were bad.

If I am doing my job OK, that VM will never once be booted in its years-long lifetime. But
if  the boss ever sees that "TCP performance may be compromised" message on a
production server, or if it ever kills or cripples the OS or apps, I will have a very personal
problem very rapidly.

It's really more about sociology than about computer science.
 

--
Sigmentation fault. Core dumped.




Reply to: