7:11 p.m.
While troubleshooting a fresh installation of (after a failed one) that caused all the
hosts but the one running the hosted-engine to become in “Unassigned” state I noticed that
the ovirt-engine complains about not being able to contact the VDSM.
I noticed that VDSM has stopped listening on IPv4.
I didn’t disable any IPv6 as it states not to disable it on hosts that are capable running
the hosted-engine and it seems that the reason behind it is that the hosted-engine talks
to the host it runs on through “localhost”, this also explains why the host which the
hosted-engine runs on is “OK”.
Below is from a host that does not run the hosted-engine:
# ss -atn | grep 543
LISTEN 0 5 *:54322 *:*
ESTAB 0 0 127.0.0.1:54792 127.0.0.1:54321
ESTAB 0 0 127.0.0.1:54798 127.0.0.1:54321
LISTEN 0 5 [::]:54321 [::]:*
ESTAB 0 0 [::ffff:127.0.0.1]:54321 [::ffff:127.0.0.1]:54798
ESTAB 0 0 [::ffff:127.0.0.1]:54321 [::ffff:127.0.0.1]:54792
ESTAB 0 0 [::1]:54321 [::1]:50238
ESTAB 0 0 [::1]:50238 [::1]:54321
Below is from a host that runs the hosted-engine at the moment:
# ss -atn | grep 543
LISTEN 0 5 *:54322 *:*
LISTEN 0 5 [::]:54321 [::]:*
ESTAB 0 0 [::1]:51230 [::1]:54321
ESTAB 0 0 [::1]:54321 [::1]:51242
ESTAB 0 0 [::ffff:10.46.20.23]:54321
[::ffff:10.46.20.20]:45706
ESTAB 0 0 [::ffff:10.46.20.23]:54321
[::ffff:10.46.20.20]:45746
ESTAB 0 0 [::1]:51240 [::1]:54321
ESTAB 0 0 [::1]:54321 [::1]:51230
ESTAB 0 0 [::1]:51242 [::1]:54321
ESTAB 0 0 [::1]:54321 [::1]:51240
The hosted-engine IP is 10.46.20.20 and the host is 10.46.20.23.
/etc/hosts on all hosts:
127.0.0.1 localhost localhost.localdomain localhost4 localhost4.localdomain4
::1 localhost localhost.localdomain localhost6 localhost6.localdomain6
Perhaps this is relevant but all hosts are enrolled into IDM (FreeIPA) and as an outcome
they all have a DNS record and a PTR record as well as the ovirt-engine VM.
# cat /etc/vdsm/vdsm.conf
[vars]
ssl = true
ssl_ciphers = HIGH:!aNULL
ssl_excludes = OP_NO_TLSv1,OP_NO_TLSv1_1
[addresses]
management_port = 54321
I have tried adding “management_ip = 0.0.0.0” but then it only binds to IPv4 and yet, the
host still shows as Unassigned, sometimes it switches to “NonResponsive” and trying to
“Reinstall” the host fails, the ovirt-engine complains it can't contact/reach the
VDSM, while using netcat from the ovirt-engine it works.
I have KSM and Memory Ballooning enabled on the Cluster as well.
oVirt 4.3.10 installed on CentOS 7.8.2003
The self-hosted Engine runs on an external GlusterFS, before reinstalling everything
(fresh start of OS, etc..) I tried iSCSI as well.