Dear all,
I've been using ovirt for at least 6 years and only lately I've
stepped into a weird problem that I hope someone will be able to
give help.
My hardware is:
- blade lenovo for hosts, with dual switch
- dell equallogic for iscsi storage, directly connected to the blade
switches
- The two host network cards are configured with bonding and all the
vlans are accessed from it (mtu 9000)
- all the hosts and ovirt engine have firewalld service disabled)
My engine is hosted on a separate vmware vm (I will evaluate the
self hosted engine later...). I want to stress the fact that for
years this setup worked smoothly without any significant issue (and
all the minor updates were completed flawlessly).
A few weeks ago I started the update from the rock solid 4.3 to the
latest 4.4.7. I began with the manager, following the docs,
installing a new centos8 vm and importing the backup: everything
went smootly and I was able to get access to the manager without any
problem, all the machines still there :)
I then began updating the hosts, from centos7 to centos8 stream, one
by one.
Immediately I noticed network issues, with the VMs hosted on the
first updated host. Migrating VMs from centos8 host to other centos8
quite often fails, but the main issue is this: if I start one of
the VMs on the centos8 host, they have no network connectivity. If
I migrate them to a centos7 hosts the network starts to work, and
if I migrate the VMs back to the centos8 host, the network keeps
working.
I am puzzled and can't understand what's going on. Generally
speaking all the centos8 hosts (I have 6 in my cluster, and now 3
are centos8 while the rest is still centos7) seem to be very
unstable, meaning that the VMs they host are quite often showing
network issues and temporary glitches.
Can someone give a hint on how to solve this weird issue?
Thanks,
Andrea
--
Andrea Chierici - INFN-CNAF
Viale Berti Pichat 6/2, 40127 BOLOGNA
Office Tel: +39 051 2095463
SkypeID ataruz
--