Ciao Marco, anzitutto grazie!
On 05/11/22 22:00, Marco Gaiarin wrote:
[...]
Se non hai quorum (stile: hai tre nodi, quorum a 2 (il default) e un
nodo
fermo) è normale che le macchine non partano.
dunque, forse non sono stato chiaro... i 3 nodi vanno tutti e 3. Mi
posso connettere ad ognuno dei 3 nodi anche tramite interfaccia web
ma da un nodo non si vedono gli altri 2. È il cluster che ha dei
problemi. Il servizio che non riesce a partire è il "Proxmox VE
replication runner" (pvesr), fra le altre cose non ho alcuna replica
impostata a memoria, utilizzo solo ceph e mi sono liberato di zfs.
Tutte gli host virtuali funzionano bene non hanno problemi tranne che
non funziona più il backup e non funzionando il cluster non posso
nemmeno migrarli di nodo; inoltre se spengo una macchina poi non
riesco più a riaccenderla.
'pvesr' è il servizio di replicazione degli storage ZFS; se non hai
ZFS o
non usi la replicazione, quel'errore è ininfluente (e comunque nulla
centra
con il quorum).
A naso hai corosync fermo su un nodo, e magai ti basta riavviare
quello (o
riavviare corosync su quello); prova a lanciare:
pvecm status
su tutti i nodi e vedi che ti dice.
Corosync non ha problemi, l'ho riavviato su tutti e 3 i nodi e non ha
dato problemi. Questo è l'output di pvecm status:
# pvecm status
Cluster information
-------------------
Name: CSA-cluster1
Config Version: 3
Transport: knet
Secure auth: on
Quorum information
------------------
Date: Mon Nov 7 08:11:51 2022
Quorum provider: corosync_votequorum
Nodes: 1
Node ID: 0x00000001
Ring ID: 1.91e
Quorate: No
Votequorum information
----------------------
Expected votes: 3
Highest expected: 3
Total votes: 1
Quorum: 2 Activity blocked
Flags:
Membership information
----------------------
Nodeid Votes Name
0x00000001 1 192.168.255.1 (local)
In syslog è pieno di messaggi tipo:
Nov 7 08:14:33 pve01 pveproxy[2699797]: Cluster not quorate -
extending auth key lifetime!
Nov 7 08:14:35 pve01 pveproxy[2699797]: Cluster not quorate -
extending auth key lifetime!
Inoltre Tutto è iniziato venerdì sera e in syslog ho trovato:
4 23:37:01 pve02 systemd[1]: Started Proxmox VE replication runner.
Nov 4 23:37:01 pve02 CRON[2145590]: (root) CMD (if test -x
/usr/sbin/apticron; then /usr/sbin/apticron --cron; else true; fi)
Nov 4 23:38:00 pve02 systemd[1]: Starting Proxmox VE replication
runner...
Nov 4 23:38:01 pve02 systemd[1]: pvesr.service: Succeeded.
Nov 4 23:38:01 pve02 systemd[1]: Started Proxmox VE replication runner.
Nov 4 23:38:26 pve02 corosync[1703]: [KNET ] link: host: 3 link:
0 is down
Nov 4 23:38:26 pve02 corosync[1703]: [KNET ] host: host: 3
(passive) best link: 0 (pri: 1)
Nov 4 23:38:26 pve02 corosync[1703]: [KNET ] host: host: 3 has no
active links
Nov 4 23:38:28 pve02 corosync[1703]: [TOTEM ] Token has not been
received in 2737 ms
Nov 4 23:38:30 pve02 corosync[1703]: [KNET ] rx: host: 3 link: 0
is up
Nov 4 23:38:30 pve02 corosync[1703]: [KNET ] host: host: 3
(passive) best link: 0 (pri: 1)
Nov 4 23:38:32 pve02 corosync[1703]: [QUORUM] Sync members[2]: 1 2
Nov 4 23:38:32 pve02 corosync[1703]: [QUORUM] Sync left[1]: 3
Nov 4 23:38:32 pve02 corosync[1703]: [TOTEM ] A new membership
(1.873) was formed. Members left: 3
Nov 4 23:38:32 pve02 corosync[1703]: [TOTEM ] Failed to receive
the leave message. failed: 3
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice: members: 1/1626,
2/1578
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice: starting data
syncronisation
Nov 4 23:38:32 pve02 pmxcfs[1578]: [status] notice: members: 1/1626,
2/1578
Nov 4 23:38:32 pve02 pmxcfs[1578]: [status] notice: starting data
syncronisation
Nov 4 23:38:32 pve02 corosync[1703]: [QUORUM] Members[2]: 1 2
Nov 4 23:38:32 pve02 corosync[1703]: [MAIN ] Completed service
synchronization, ready to provide service.
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice: received sync
request (epoch 1/1626/00000009)
Nov 4 23:38:32 pve02 pmxcfs[1578]: [status] notice: received sync
request (epoch 1/1626/00000009)
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice: received all states
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice: leader is 1/1626
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice: synced members:
1/1626, 2/1578
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice: all data is up to
date
Nov 4 23:38:32 pve02 pmxcfs[1578]: [dcdb] notice:
dfsm_deliver_queue: queue length 2
Nov 4 23:38:32 pve02 pmxcfs[1578]: [status] notice: received all states
Nov 4 23:38:32 pve02 pmxcfs[1578]: [status] notice: all data is up
to date
Nov 4 23:38:32 pve02 pmxcfs[1578]: [status] notice:
dfsm_deliver_queue: queue length 46
Nov 4 23:38:34 pve02 corosync[1703]: [KNET ] link: host: 3 link:
0 is down
Nov 4 23:38:34 pve02 corosync[1703]: [KNET ] host: host: 3
(passive) best link: 0 (pri: 1)
Nov 4 23:38:34 pve02 corosync[1703]: [KNET ] host: host: 3 has no
active links
Nov 4 23:38:41 pve02 corosync[1703]: [KNET ] link: host: 1 link:
0 is down
Nov 4 23:38:41 pve02 corosync[1703]: [KNET ] host: host: 1
(passive) best link: 0 (pri: 1)
Nov 4 23:38:41 pve02 corosync[1703]: [KNET ] host: host: 1 has no
active links
Nov 4 23:38:42 pve02 corosync[1703]: [TOTEM ] Token has not been
received in 2737 ms
Nov 4 23:38:43 pve02 corosync[1703]: [TOTEM ] A processor failed,
forming new configuration: token timed out (3650ms), waiting 4380ms
for consensus.
Nov 4 23:38:48 pve02 corosync[1703]: [QUORUM] Sync members[1]: 2
Nov 4 23:38:48 pve02 corosync[1703]: [QUORUM] Sync left[1]: 1
Nov 4 23:38:48 pve02 corosync[1703]: [TOTEM ] A new membership
(2.877) was formed. Members left: 1
Nov 4 23:38:48 pve02 corosync[1703]: [TOTEM ] Failed to receive
the leave message. failed: 1
Nov 4 23:38:48 pve02 pmxcfs[1578]: [dcdb] notice: members: 2/1578
Nov 4 23:38:48 pve02 pmxcfs[1578]: [status] notice: members: 2/1578
Nov 4 23:38:48 pve02 corosync[1703]: [QUORUM] This node is within
the non-primary component and will NOT provide any services.
Nov 4 23:38:48 pve02 corosync[1703]: [QUORUM] Members[1]: 2
Nov 4 23:38:48 pve02 corosync[1703]: [MAIN ] Completed service
synchronization, ready to provide service.
Nov 4 23:38:48 pve02 pmxcfs[1578]: [status] notice: node lost quorum
Nov 4 23:38:48 pve02 pmxcfs[1578]: [dcdb] crit: received write while
not quorate - trigger resync
Nov 4 23:38:48 pve02 pmxcfs[1578]: [dcdb] crit: leaving CPG group
Nov 4 23:38:48 pve02 pve-ha-lrm[1943]: unable to write lrm status
file - unable to open file '/etc/pve/nodes/pve02/lrm_status.tmp.1943'
- Permission denied
Nov 4 23:38:49 pve02 pmxcfs[1578]: [dcdb] notice: start cluster
connection
Nov 4 23:38:49 pve02 pmxcfs[1578]: [dcdb] crit: cpg_join failed: 14
Nov 4 23:38:49 pve02 pmxcfs[1578]: [dcdb] crit: can't initialize
service
Nov 4 23:38:55 pve02 pmxcfs[1578]: [dcdb] notice: members: 2/1578
Nov 4 23:38:55 pve02 pmxcfs[1578]: [dcdb] notice: all data is up to
date
Nov 4 23:39:00 pve02 systemd[1]: Starting Proxmox VE replication
runner...
Nov 4 23:39:01 pve02 pvesr[2146320]: trying to acquire cfs lock
'file-replication_cfg' ...
[...]
Cercando in rete ho visto che non sono il solo ad avere questo
problema e tutti indirizzano ad impostare il quorum a uno con pvecm
expected 1 ma sono un po' preoccupato prima di fare qualunque cosa
vorrei esserne ipercerto non avendo nemmeno il backup delle macchine
virtuali!
Grazie un monte
Piviul