[Ticket#:20737315] Fw: slowness of profitbricks-build9-amd64.debian.net aka 46.16.78.222 C873536754

Helmut Grohne helmut at subdivi.de
Sun Oct 20 06:44:22 BST 2019


Hi,

On Thu, Oct 17, 2019 at 09:08:24AM +0200, IONOS Enterprise Support wrote:
> first, please make your test again now.

In the openssl test, the machine appears around 6 times faster than
previously now. The performance numbers from openssl speed are
reasonable and expected now. I've triggered the regular CI workload now
and the performance looks much better.

> Second, for performance measures we use iperf and fio. So I would have to
> check if openssl is a candidat as performance tool for the CPU. To shorten
> this up you may switch to Intel. What happens then?

I only used openssl here, because it was so readily available and
because it seemes remotely similar to the target workload. I'd have to
install something else. I didn't see a use to have a very precise
benchmark here, because it was fairly obvious that the performance
numbers were wrong by a magnitude.

Would you mind telling what changed about the machine? In any case, we
can mark the ticket as resolved and skip futher research.

Thank you

Helmut



More information about the Qa-jenkins-dev mailing list