Distribution EOLE

• Les 3 VM hapy1, hapy2 et hapy3 ont la même adresse IP, du coup les requêtes réseaux n'aboutissent pas correctement
• Si ont désactives les 2 autres VM, les commandes host hapy1, host hapy2 et host hapy3 fonctionnent
  • Toutefois l'interface ens4 est à l'état down au moment de la génération des règles de firewall, donc pas de requête réseau possible. Si j'ajoute ip link set ens4 up juste avant la ligne . /usr/share/eole/firewall.start de posttemplate/00-eole-common, l'instance passe.

Trouvé : si openvswitch-switch n'est pas démarré :

root@hapy2:~# systemctl stop openvswitch-switch.service
root@hapy2:~# systemctl restart netplan-apply.service
Job for netplan-apply.service failed because the control process exited with error code.
See "systemctl status netplan-apply.service" and "journalctl -xe" for details.
root@hapy2:~# systemctl status netplan-apply.service
● netplan-apply.service - Apply netplan config
     Loaded: loaded (/lib/systemd/system/netplan-apply.service; enabled; vendor preset: enabled)
     Active: failed (Result: exit-code) since Fri 2020-11-20 15:39:16 CET; 1min 1s ago
    Process: 38658 ExecStart=/usr/sbin/netplan apply (code=exited, status=1/FAILURE)
   Main PID: 38658 (code=exited, status=1/FAILURE)

nov. 20 15:39:16 hapy2 netplan[38658]:   File "/usr/share/netplan/netplan/cli/ovs.py", line 128, in apply_ovs_cleanup
nov. 20 15:39:16 hapy2 netplan[38658]:     out = subprocess.check_output([OPENVSWITCH_OVS_VSCTL, '--columns=name,external-ids',
nov. 20 15:39:16 hapy2 netplan[38658]:   File "/usr/lib/python3.8/subprocess.py", line 411, in check_output
nov. 20 15:39:16 hapy2 netplan[38658]:     return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
nov. 20 15:39:16 hapy2 netplan[38658]:   File "/usr/lib/python3.8/subprocess.py", line 512, in run
nov. 20 15:39:16 hapy2 netplan[38658]:     raise CalledProcessError(retcode, process.args,
nov. 20 15:39:16 hapy2 netplan[38658]: subprocess.CalledProcessError: Command '['/usr/bin/ovs-vsctl', '--columns=name,external-ids', '-f', 'csv', '-d', 'bare', '--no-headings', 'list', 'Port']' returned non-zero exit status 1.
nov. 20 15:39:16 hapy2 systemd[1]: netplan-apply.service: Main process exited, code=exited, status=1/FAILURE
nov. 20 15:39:16 hapy2 systemd[1]: netplan-apply.service: Failed with result 'exit-code'.
nov. 20 15:39:16 hapy2 systemd[1]: Failed to start Apply netplan config.

Le problème n'est pas lié au mode HA. Il existe également sur instance-default-2.8.0 : https://dev-eole.ac-dijon.fr/jenkins/job/2.8.0/job/test-instance-acahapy-2.8.0-amd64/202/console
Ce problème n'existe pas en 2.7.2

https://dev-eole.ac-dijon.fr/squash/executions/11328

reconfigure hapy2 ou 3 :

run-parts: executing /usr/share/eole/postservice/99-one-hooks-manager reconfigure
[one.hook.update] Cannot process request, no leader found
Warning: vm-close_ports hook creation failed !
[one.hook.update] Cannot process request, no leader found
Warning: vm-close_ports_on_shutdown hook creation failed !

reconfigure hapy1, Erreur en postservice/30-one-mng :

Host default already present in cluster
[ERROR] [Host sync failed]

Tests repassés avec succès le 2/12/2020
il subsiste une erreur en postservice/30-one-mng

parts: executing /usr/share/eole/postservice/29-ovs-mng reconfigure
Open vSwitch configuration
   Virtual switch vswitch already exist
        Cleaning port for interface ens4
        Cleaning port for interface ens5
        Cleaning port aca
   Port aca created
run-parts: executing /usr/share/eole/postservice/30-one-mng reconfigure
Wait for oned to be started.
The default cluster name is already "default" 
Host default already present in cluster
[ERROR][Hosts sync failed]
run-parts: /usr/share/eole/postservice/30-one-mng exited with return code 1
Erreur : postservice

Mais le cluster est fonctionnel selon la commande onehost list.
L'authentification web fonctionne aussi.