[pve-devel] [PATCH V2 pve-ha-manager] Fix shutdown order of ceph daemons.

[Wolfgang Link]

On 04/26/2017 02:19 PM, Thomas Lamprecht wrote:
> On 04/26/2017 01:17 PM, Wolfgang Link wrote:
>> It is important that storages stop after pve-ha-lrm.
>>
>> If the storages  stops to early the vm loses disks
>> and can not shutdown.
>>
>> This can ends in fencing the node.
>> ---
>>   debian/pve-ha-lrm.service | 1 +
>>   1 file changed, 1 insertion(+)
>>
>> diff --git a/debian/pve-ha-lrm.service b/debian/pve-ha-lrm.service
>> index 3dc2ebc..8da0eb4 100644
>> --- a/debian/pve-ha-lrm.service
>> +++ b/debian/pve-ha-lrm.service
>> @@ -6,6 +6,7 @@ Wants=watchdog-mux.service
>>   Wants=pvedaemon.service
>>   Wants=pve-ha-crm.service
>>   Wants=lxc.service
>> +After=pve-storage.target
> I'd also add the
> 
> Wants=pve-storage.target
> 
> To ensure that all storages are up when the HA stack begins to start
> VMs/CTs?
IMHO the wants is not necessary because

pve-ha-lrm and pveproxy is wanted by multi-user.target

pveproxy wants pve-storage.

Wants will only start the service.
But in this case pveproxy does the job.
So "after" should bring the correct order.


> Else acked by me.
> 
>>   After=pve-ha-crm.service
>>   After=pve-cluster.service
>>   After=corosync.service
> 
> 
> _______________________________________________
> pve-devel mailing list
> pve-devel at pve.proxmox.com
> http://pve.proxmox.com/cgi-bin/mailman/listinfo/pve-devel