[ovirt-users] Re: Lots of problems with deploying the hosted-engine (ovirt 4.4 | CentOS 8.2.2004)

2020-07-13 Thread Sandro Bonazzola
Il giorno dom 28 giu 2020 alle ore 16:38 jonas  ha
scritto:

> Hi!
>
> I have banged my head against deploying the ovirt 4.4 self-hosted engine
> on Centos 8.2 for last couple of days.
>
> First I was astonished that resources.ovirt.org has no IPv6
> connectivity, which made my initial plan for a mostly IPv6-only
> deployment impossible.
>


We have an open ticket about it here:
https://ovirt-jira.atlassian.net/browse/OVIRT-2268
You can try using one of the mirrors:
https://resources.ovirt.org/pub/yum-repo/mirrorlist
One of them may have IPv6 enabled.


>
> CentOS was installed from scratch using the ks.cgf Kickstart file below,
> which also adds the ovirt 4.4 repo and installs cockpit-ovirt-dashboard
> & ovirt-engine-appliance.
>
> When deploying the hosted-engine from cockpit while logged in as a
> non-root (although privileged) user, the "(3) Prepare VM" step instantly
> fails with a nondescript error message and without generating any logs.
> By using the browser dev tools it was determined that this was because
> the ansible vars file could not be created as the non-root user did not
> have write permissions in '/var/lib/ovirt-hosted-engine-setup/cockpit/'
> . Shouldn't cockpit be capable of using sudo when appropriate, or at
> least give a more descriptive error message?
>

oVirt doesn't support running hosted engine deployment as non-root user.


>
> After login into cockpit as root, or when using the command line
> ovirt-hosted-engine-setup tool, the deployment fails with "Failed to
> download metadata for repo 'AppStream'".
> This seems to be because a) the dnsmasq running on the host does not
> forward dns queries, even though the host itself can resolve dns queries
> just fine, and b) there also does not seem to be any functioning routing
> setup to reach anything outside the host.
> Regarding a) it is strange that dnsmasq is running with a config file
> '/var/lib/libvirt/dnsmasq/default.conf' containing the 'no-resolv'
> option. Could the operation of systemd-resolved be interfering with
> dnsmasq (see ss -tulpen output)? I tried to manually stop
> systemd-resolved, but got the same behaviour as before.
>

Never seen something like this happening, +Dominik Holler
 can you help here?


>
> I hope someone could give me a hint how I could get past this problem,
> as so far my ovirt experience has been a little bit sub-par. :D
>
> Also when running ovirt-hosted-engine-cleanup, the extracted engine VMs
> in /var/tmp/localvm* are not removed, leading to a "disk-memory-leak"
> with subsequent runs.
>

yes the recomendation is to run again the deployment from scratch, the
cleanup tool has no guarantee to remove everything from the host so it may
have some file left around on the system.



>
> Best regards
> Jonas
>
> --- ss -tulpen output post deploy-run ---
> [root@nxtvirt ~]# ss -tulpen  | grep ':53 '
> udp   UNCONN 0  0 127.0.0.53%lo:53
> 0.0.0.0:* users:(("systemd-resolve",pid=1379,fd=18)) uid:193
> ino:32910 sk:6 <->
> udp   UNCONN 0  0   [fd00:1234:5678:900::1]:53
> [::]:* users:(("dnsmasq",pid=13525,fd=15)) uid:979 ino:113580
> sk:d v6only:1 <->
> udp   UNCONN 0  0  [fe80::5054:ff:fe94:f314]%virbr0:53
> [::]:* users:(("dnsmasq",pid=13525,fd=12)) uid:979 ino:113575
> sk:e v6only:1 <->
> tcp   LISTEN 0  32  [fd00:1234:5678:900::1]:53
> [::]:* users:(("dnsmasq",pid=13525,fd=16)) uid:979 ino:113581
> sk:20 v6only:1 <->
> tcp   LISTEN 0  32 [fe80::5054:ff:fe94:f314]%virbr0:53
> [::]:* users:(("dnsmasq",pid=13525,fd=13)) uid:979 ino:113576
> sk:21 v6only:1 <->
>
>
> --- running dnsmasq processes on host ('nxtvirt') post deploy-run ---
>
> dnsmasq13525  0.0  0.0  71888  2344 ?S12:31   0:00
> /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf
> --leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper
> root   13526  0.0  0.0  71860   436 ?S12:31   0:00
> /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf
> --leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper
>
>
> --- var/lib/libvirt/dnsmasq/default.conf ---
>
> ##WARNING:  THIS IS AN AUTO-GENERATED FILE. CHANGES TO IT ARE LIKELY TO
> BE
> ##OVERWRITTEN AND LOST.  Changes to this configuration should be made
> using:
> ##virsh net-edit default
> ## or other application using the libvirt API.
> ##
> ## dnsmasq conf file created by libvirt
> strict-order
> pid-file=/run/libvirt/network/default.pid
> except-interface=lo
> bind-dynamic
> interface=virbr0
> dhcp-option=3
> no-resolv
> ra-param=*,0,0
> dhcp-range=fd00:1234:5678:900::10,fd00:1234:5678:900::ff,64
> dhcp-lease-max=240
> dhcp-hostsfile=/var/lib/libvirt/dnsmasq/default.hostsfile
> addn-hosts=/var/lib/libvirt/dnsmasq/default.addnhosts
> enable-ra
>
> --- cockpit wizard overview before the 'Prepare VM' step ---
>
> VM
>  Engine FQDN:engine.*REDACTED*
>  MAC Address:00:16:3e:20:13:b3
>  

[ovirt-users] Re: Lots of problems with deploying the hosted-engine (ovirt 4.4 | CentOS 8.2.2004)

2020-07-08 Thread Asaf Rachmani
Hi Jonas,

Can you please share the full log file located in
/var/log/ovirt-hosted-engine-setup ?
My guess you hit [1], this is the fix for that [2] (not merged yet).
You can install the rpm with the fix from Jenkins CI [3] and then run the
following command:
hosted-engine --deploy --ansible-extra-vars=he_offline_deployment=true

Regarding the "disk-memory-leak", can you please open a bug for that?

Regards,
Asaf

[1] https://bugzilla.redhat.com/show_bug.cgi?id=1849517
[2] https://gerrit.ovirt.org/109730
[3]
https://jenkins.ovirt.org/job/ovirt-hosted-engine-setup_standard-check-patch/184/


On Tue, Jul 7, 2020 at 4:24 PM jonas  wrote:

> Hi!
>
> A short addendum:
> I have now also tried to perform the installation using the oVirt Node
> distribution as a basis, but that also ended with the same problem. So
> it does not seam to be an issue with the underlying CentOS installation,
> but rather with my general setup or parameters.
>
> Regards
> Jonas
>
> On 2020-06-28 16:26, jonas wrote:
> > Hi!
> >
> > I have banged my head against deploying the ovirt 4.4 self-hosted
> > engine on Centos 8.2 for last couple of days.
> >
> > First I was astonished that resources.ovirt.org has no IPv6
> > connectivity, which made my initial plan for a mostly IPv6-only
> > deployment impossible.
> >
> > CentOS was installed from scratch using the ks.cgf Kickstart file
> > below, which also adds the ovirt 4.4 repo and installs
> > cockpit-ovirt-dashboard & ovirt-engine-appliance.
> >
> > When deploying the hosted-engine from cockpit while logged in as a
> > non-root (although privileged) user, the "(3) Prepare VM" step
> > instantly fails with a nondescript error message and without
> > generating any logs. By using the browser dev tools it was determined
> > that this was because the ansible vars file could not be created as
> > the non-root user did not have write permissions in
> > '/var/lib/ovirt-hosted-engine-setup/cockpit/' . Shouldn't cockpit be
> > capable of using sudo when appropriate, or at least give a more
> > descriptive error message?
> >
> > After login into cockpit as root, or when using the command line
> > ovirt-hosted-engine-setup tool, the deployment fails with "Failed to
> > download metadata for repo 'AppStream'".
> > This seems to be because a) the dnsmasq running on the host does not
> > forward dns queries, even though the host itself can resolve dns
> > queries just fine, and b) there also does not seem to be any
> > functioning routing setup to reach anything outside the host.
> > Regarding a) it is strange that dnsmasq is running with a config file
> > '/var/lib/libvirt/dnsmasq/default.conf' containing the 'no-resolv'
> > option. Could the operation of systemd-resolved be interfering with
> > dnsmasq (see ss -tulpen output)? I tried to manually stop
> > systemd-resolved, but got the same behaviour as before.
> >
> > I hope someone could give me a hint how I could get past this problem,
> > as so far my ovirt experience has been a little bit sub-par. :D
> >
> > Also when running ovirt-hosted-engine-cleanup, the extracted engine
> > VMs in /var/tmp/localvm* are not removed, leading to a
> > "disk-memory-leak" with subsequent runs.
> >
> > Best regards
> > Jonas
> >
> > --- ss -tulpen output post deploy-run ---
> > [root@nxtvirt ~]# ss -tulpen  | grep ':53 '
> > udp   UNCONN 0  0 127.0.0.53%lo:53
> > 0.0.0.0:* users:(("systemd-resolve",pid=1379,fd=18)) uid:193
> > ino:32910 sk:6 <->
> > udp   UNCONN 0  0   [fd00:1234:5678:900::1]:53
> > [::]:* users:(("dnsmasq",pid=13525,fd=15)) uid:979 ino:113580
> > sk:d v6only:1 <->
> > udp   UNCONN 0  0  [fe80::5054:ff:fe94:f314]%virbr0:53
> > [::]:* users:(("dnsmasq",pid=13525,fd=12)) uid:979 ino:113575
> > sk:e v6only:1 <->
> > tcp   LISTEN 0  32  [fd00:1234:5678:900::1]:53
> > [::]:* users:(("dnsmasq",pid=13525,fd=16)) uid:979 ino:113581
> > sk:20 v6only:1 <->
> > tcp   LISTEN 0  32 [fe80::5054:ff:fe94:f314]%virbr0:53
> > [::]:* users:(("dnsmasq",pid=13525,fd=13)) uid:979 ino:113576
> > sk:21 v6only:1 <->
> >
> >
> > --- running dnsmasq processes on host ('nxtvirt') post deploy-run ---
> >
> > dnsmasq13525  0.0  0.0  71888  2344 ?S12:31   0:00
> > /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf
> > --leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper
> > root   13526  0.0  0.0  71860   436 ?S12:31   0:00
> > /usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf
> > --leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper
> >
> >
> > --- var/lib/libvirt/dnsmasq/default.conf ---
> >
> > ##WARNING:  THIS IS AN AUTO-GENERATED FILE. CHANGES TO IT ARE LIKELY TO
> > BE
> > ##OVERWRITTEN AND LOST.  Changes to this configuration should be made
> > using:
> > ##virsh net-edit default
> > ## or other application using the libvirt API.
> > ##
> > ## dnsmasq conf file created 

[ovirt-users] Re: Lots of problems with deploying the hosted-engine (ovirt 4.4 | CentOS 8.2.2004)

2020-07-07 Thread shadow emy
I am using command line hosted-engine --deploy for install, no cockpit.
I had problems with rpm metadata and the deployed failed, but yes in your case 
might be IPv6 problems.
I am using IPv4, never tried IPv6 on Ovirt.
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/privacy-policy.html
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/OUWKGGVZP2UN7NZPRTNXJ6WJ3ONK36QO/


[ovirt-users] Re: Lots of problems with deploying the hosted-engine (ovirt 4.4 | CentOS 8.2.2004)

2020-07-07 Thread shadow emy
Yes i also had a lot of problems installing ovirt 4.4 .I think it was not 
tested enough.
I am upgrading  from ovirt 4.3 to 4.4 using shared storage glusterfs, which 
makes things more difficult.

Regarding your error, i believe is something with the rpm ovirt 4.4 
repository(sometimes it times out and sometimes it's not).
First as you said check your disk space, before running the hosted-engine 
deploy.If the setup fails there is no ansible disk cleanup  task for /var/tmp, 
the clean task is only at the end of the ansible playbook when the deploy is 
with success.
What i tried with success  is to clear the rpm packages metadata, just before 
the deploy started  run : "dnf clean all" , then "dnf update" .
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/privacy-policy.html
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/DDSE6MI7XCB6T75ARSRFXY3FCE62TECU/


[ovirt-users] Re: Lots of problems with deploying the hosted-engine (ovirt 4.4 | CentOS 8.2.2004)

2020-07-07 Thread jonas
Hi!
Thanks for your comments! I am unsure regarding your hunch that the oVirt Repo 
is to blame, as from inside the engine after the botched deployment, absolutely 
nothing outside of the engine and its host is reachable via IPv4/v6, and a 
manual "dnf update" also produces a failure that the AppStream metadata can't 
be fetched.  Therefore I still believe it's a bug with he temporary 
networking/dns setup used when deploying the hosted engine.
I have sufficient space on the host, even after the engine was (half-)deployed 
there is still 80% free space available.
The only connection problems I had with the oVirt repo were that it is not IPv6 
ready.
___
Users mailing list -- users@ovirt.org
To unsubscribe send an email to users-le...@ovirt.org
Privacy Statement: https://www.ovirt.org/privacy-policy.html
oVirt Code of Conduct: 
https://www.ovirt.org/community/about/community-guidelines/
List Archives: 
https://lists.ovirt.org/archives/list/users@ovirt.org/message/KBP3BGXOZBKP3GSHUDAANFXHGVCKFNFD/


[ovirt-users] Re: Lots of problems with deploying the hosted-engine (ovirt 4.4 | CentOS 8.2.2004)

2020-07-07 Thread jonas

Hi!

A short addendum:
I have now also tried to perform the installation using the oVirt Node 
distribution as a basis, but that also ended with the same problem. So 
it does not seam to be an issue with the underlying CentOS installation, 
but rather with my general setup or parameters.


Regards
Jonas

On 2020-06-28 16:26, jonas wrote:

Hi!

I have banged my head against deploying the ovirt 4.4 self-hosted
engine on Centos 8.2 for last couple of days.

First I was astonished that resources.ovirt.org has no IPv6
connectivity, which made my initial plan for a mostly IPv6-only
deployment impossible.

CentOS was installed from scratch using the ks.cgf Kickstart file
below, which also adds the ovirt 4.4 repo and installs
cockpit-ovirt-dashboard & ovirt-engine-appliance.

When deploying the hosted-engine from cockpit while logged in as a
non-root (although privileged) user, the "(3) Prepare VM" step
instantly fails with a nondescript error message and without
generating any logs. By using the browser dev tools it was determined
that this was because the ansible vars file could not be created as
the non-root user did not have write permissions in
'/var/lib/ovirt-hosted-engine-setup/cockpit/' . Shouldn't cockpit be
capable of using sudo when appropriate, or at least give a more
descriptive error message?

After login into cockpit as root, or when using the command line
ovirt-hosted-engine-setup tool, the deployment fails with "Failed to
download metadata for repo 'AppStream'".
This seems to be because a) the dnsmasq running on the host does not
forward dns queries, even though the host itself can resolve dns
queries just fine, and b) there also does not seem to be any
functioning routing setup to reach anything outside the host.
Regarding a) it is strange that dnsmasq is running with a config file
'/var/lib/libvirt/dnsmasq/default.conf' containing the 'no-resolv'
option. Could the operation of systemd-resolved be interfering with
dnsmasq (see ss -tulpen output)? I tried to manually stop
systemd-resolved, but got the same behaviour as before.

I hope someone could give me a hint how I could get past this problem,
as so far my ovirt experience has been a little bit sub-par. :D

Also when running ovirt-hosted-engine-cleanup, the extracted engine
VMs in /var/tmp/localvm* are not removed, leading to a
"disk-memory-leak" with subsequent runs.

Best regards
Jonas

--- ss -tulpen output post deploy-run ---
[root@nxtvirt ~]# ss -tulpen  | grep ':53 '
udp   UNCONN 0  0 127.0.0.53%lo:53
0.0.0.0:* users:(("systemd-resolve",pid=1379,fd=18)) uid:193
ino:32910 sk:6 <->
udp   UNCONN 0  0   [fd00:1234:5678:900::1]:53
[::]:* users:(("dnsmasq",pid=13525,fd=15)) uid:979 ino:113580
sk:d v6only:1 <->
udp   UNCONN 0  0  [fe80::5054:ff:fe94:f314]%virbr0:53
[::]:* users:(("dnsmasq",pid=13525,fd=12)) uid:979 ino:113575
sk:e v6only:1 <->
tcp   LISTEN 0  32  [fd00:1234:5678:900::1]:53
[::]:* users:(("dnsmasq",pid=13525,fd=16)) uid:979 ino:113581
sk:20 v6only:1 <->
tcp   LISTEN 0  32 [fe80::5054:ff:fe94:f314]%virbr0:53
[::]:* users:(("dnsmasq",pid=13525,fd=13)) uid:979 ino:113576
sk:21 v6only:1 <->


--- running dnsmasq processes on host ('nxtvirt') post deploy-run ---

dnsmasq13525  0.0  0.0  71888  2344 ?S12:31   0:00
/usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf
--leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper
root   13526  0.0  0.0  71860   436 ?S12:31   0:00
/usr/sbin/dnsmasq --conf-file=/var/lib/libvirt/dnsmasq/default.conf
--leasefile-ro --dhcp-script=/usr/libexec/libvirt_leaseshelper


--- var/lib/libvirt/dnsmasq/default.conf ---

##WARNING:  THIS IS AN AUTO-GENERATED FILE. CHANGES TO IT ARE LIKELY TO 
BE
##OVERWRITTEN AND LOST.  Changes to this configuration should be made 
using:

##virsh net-edit default
## or other application using the libvirt API.
##
## dnsmasq conf file created by libvirt
strict-order
pid-file=/run/libvirt/network/default.pid
except-interface=lo
bind-dynamic
interface=virbr0
dhcp-option=3
no-resolv
ra-param=*,0,0
dhcp-range=fd00:1234:5678:900::10,fd00:1234:5678:900::ff,64
dhcp-lease-max=240
dhcp-hostsfile=/var/lib/libvirt/dnsmasq/default.hostsfile
addn-hosts=/var/lib/libvirt/dnsmasq/default.addnhosts
enable-ra

--- cockpit wizard overview before the 'Prepare VM' step ---

VM
Engine FQDN:engine.*REDACTED*
MAC Address:00:16:3e:20:13:b3
Network Configuration:Static
VM IP Address:*REDACTED*:1099:babe::3/64
Gateway Address:*REDACTED*:1099::1
DNS Servers:*REDACTED*:1052::11
Root User SSH Access:yes
Number of Virtual CPUs:4
Memory Size (MiB):4096
Root User SSH Public Key:(None)
Add Lines to /etc/hosts:yes
Bridge Name:ovirtmgmt
Apply OpenSCAP profile:no
Engine
SMTP Server Name:localhost
SMTP Server Port Number:25
Sender E-Mail Address:root@localhost
Recipient