On Fri, Jul 17, 2020 at 11:04 AM Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:

On Fri, Jul 17, 2020 at 10:58 AM Gianluca Cecchi <gianluca.cecchi@gmail.com> wrote:

On Fri, Jul 17, 2020 at 10:54 AM Martin Perina <mperina@redhat.com> wrote:

Hi Gianluca,

that's very strange error, because I'm 100% sure we are using yum module with Python3 in several other roles including adding host to engine or upgrading host and so far I haven't heard any issue with ansible 2.9.10 and yum module.

Gobinda, wouldn't enforcing python interpreter version help there?

https://github.com/oVirt/ovirt-engine/blob/master/packaging/ansible-runner-service-project/project/roles/ovirt-host-deploy-facts/tasks/main.yml#L28

Regards,

Martin

I have a very clean install from 4.1.1.1 node ng iso anf I'm the third to notice that with this release.

The engine deployment is going on. Not finished yet, but to have ti go I had to modify, with the same strategy ("use: dnf" with package module and use "package" instead of "yum" and also specifying "use: dnf") in these files under /usr/share/ansible/roles:

ovirt.engine-setup/tasks/engine_setup.yml
ovirt.engine-setup/tasks/install_packages.yml

ovirt.hosted_engine_setup/tasks/install_packages.yml

ovirt.hosted_engine_setup/tasks/create_target_vm/03_hosted_engine_final_tasks.yml
ovirt.hosted_engine_setup/tasks/install_appliance.yml

Gianluca

The installation from the iso was with all default values.

The only "non standard" thing, if we want it to call this way is that before running the wizard, on the host I pre-installed the appliance package, to shorten the deploy phase hereafter.

And to do it I executed, because of habit:

yum install ovirt-engine-appliance

instead of "dnf install...", but I think this doesn't influence ansible autodetect when using "package" module or the error about python2 when using "yum" module...

Gianluca

The engine deployment failed in the phase where it tries to add the host and waits for the host to be up and if I go into the logs in

/var/log/ovirt-hosted-engine-setup/engine-logs-2020-07-17T08:30:48Z/ovirt-engine/host-deploy/

the file ovirt-host-deploy-ansible-20200717104103-novirt2.example.net-3a710f0c.log contains

020-07-17 10:41:17 CEST - fatal: [novirt2.example.net]: FAILED! => {"changed": false, "module_stderr": "/bin/sh: /usr/bin
/python2: No such file or directory\n", "module_stdout": "", "msg": "The module failed to execute correctly, you probably
need to set the interpreter.\nSee stdout/stderr for the exact error", "rc": 127}
2020-07-17 10:41:17 CEST - {
  "status" : "OK",
  "msg" : "",
  "data" : {
    "uuid" : "00f4c6a8-8423-4a2a-bfd5-f38c34f56ecf",
    "counter" : 53,
    "stdout" : "fatal: [novirt2.example.net]: FAILED! => {\"changed\": false, \"module_stderr\": \"/bin/sh: /usr/bin/pytho
n2: No such file or directory\\n\", \"module_stdout\": \"\", \"msg\": \"The module failed to execute correctly, you probab
ly need to set the interpreter.\\nSee stdout/stderr for the exact error\", \"rc\": 127}",

So I think I have to find and solve why it searches python2....

I compared on an existing 4.4.0 environment I have (hci single node installed from 4.4.0 node ng iso) and no python2 apparently there, only ansible that is at  ansible-2.9.9-1.el8.noarch instead of ansible-2.9.10-1.el8.noarch of 4.4.1.1

Possibly any wrong default about python?

Alternatives seems the same between 4.4.0 and 4.4.1.1

4.4.0

[g.cecchi@ovirt01 ~]$ alternatives --list
cifs-idmap-plugin     auto   /usr/lib64/cifs-utils/cifs_idmap_sss.so
ifup                   auto   /etc/sysconfig/network-scripts/ifup
ld                     auto   /usr/bin/ld.bfd
libnssckbi.so.x86_64   auto   /usr/lib64/pkcs11/p11-kit-trust.so
libwbclient.so.0.15-64 auto   /usr/lib64/samba/wbclient/libwbclient.so.0.15
mkisofs               auto   /usr/bin/genisoimage
mta                   auto   /usr/sbin/sendmail.postfix
nmap                   auto   /usr/bin/ncat
python                 auto   /usr/libexec/no-python
python3               auto   /usr/bin/python3.6
[g.cecchi@ovirt01 ~]$

4.4.1.1

[root@novirt2 host-deploy]# alternatives --list
cifs-idmap-plugin     auto   /usr/lib64/cifs-utils/cifs_idmap_sss.so
ifup                   auto   /etc/sysconfig/network-scripts/ifup
ld                     auto   /usr/bin/ld.bfd
libnssckbi.so.x86_64   auto   /usr/lib64/pkcs11/p11-kit-trust.so
libwbclient.so.0.15-64 auto   /usr/lib64/samba/wbclient/libwbclient.so.0.15
mkisofs               auto   /usr/bin/genisoimage
mta                   auto   /usr/sbin/sendmail.postfix
nmap                   auto   /usr/bin/ncat
python                 auto   /usr/libexec/no-python
python3               auto   /usr/bin/python3.6
[root@novirt2 host-deploy]#

Not sure where to search if not somehow a bug of ansible 2.9.10

Can I try to clean install a 4.4.1.1 host and downgrade ansible before deploy, eg running

rpm -Uvh --oldpackage ansible-2.9.9-1.el8.noarch

?

Where to get the 2.9.9 rpm to apply in case, just to exclude it?

Gianluca