Power on function

olivierlambert

This is for task destroy, not a task creation.

dsiminiuk

@livierlambert it is all I see from:

fgrep Async.host.power_on /var/log/audit.log

olivierlambert

Grep on host.power_on to see if it's better.

dsiminiuk

@livierlambert

fgrep host.power_on audit.log.1 | fgrep -v destroy | fgrep -v cancel

Jul  7 23:58:02 xcpng01 xapi: [20200708T04:58:02.118Z|audit||36922 |Async.host.power_on R:0af8abcf2f46|audit] ('trackid=68849bbe6e0a22d9b0b024c21c2b6e36' 'LOCAL_SUPERUSER' 'root' 'ALLOWED' 'ERROR:INTERNAL_ERROR: [ (Failure \"The host failed to power on.\") ]' 'API' 'host.power_on' (('host' 'xcpng02' '1139bfbc-c487-4311-8206-bbffe96ed6b1' 'OpaqueRef:426e950f-f926-490f-b598-328314aafbdc')))

Jul  8 00:58:02 xcpng01 xapi: [20200708T05:58:02.803Z|audit||42462 |Async.host.power_on R:2733002e2e58|audit] ('trackid=68849bbe6e0a22d9b0b024c21c2b6e36' 'LOCAL_SUPERUSER' 'root' 'ALLOWED' 'ERROR:INTERNAL_ERROR: [ (Failure \"The host failed to power on.\") ]' 'API' 'host.power_on' (('host' 'xcpng02' '1139bfbc-c487-4311-8206-bbffe96ed6b1' 'OpaqueRef:426e950f-f926-490f-b598-328314aafbdc')))

Jul  8 01:58:03 xcpng01 xapi: [20200708T06:58:03.488Z|audit||48005 |Async.host.power_on R:e74dd3d29556|audit] ('trackid=68849bbe6e0a22d9b0b024c21c2b6e36' 'LOCAL_SUPERUSER' 'root' 'ALLOWED' 'ERROR:INTERNAL_ERROR: [ (Failure \"The host failed to power on.\") ]' 'API' 'host.power_on' (('host' 'xcpng02' '1139bfbc-c487-4311-8206-bbffe96ed6b1' 'OpaqueRef:426e950f-f926-490f-b598-328314aafbdc')))

Jul  8 02:58:04 xcpng01 xapi: [20200708T07:58:04.229Z|audit||53358 |Async.host.power_on R:e730dacdf348|audit] ('trackid=68849bbe6e0a22d9b0b024c21c2b6e36' 'LOCAL_SUPERUSER' 'root' 'ALLOWED' 'ERROR:INTERNAL_ERROR: [ (Failure \"The host failed to power on.\") ]' 'API' 'host.power_on' (('host' 'xcpng02' '1139bfbc-c487-4311-8206-bbffe96ed6b1' 'OpaqueRef:426e950f-f926-490f-b598-328314aafbdc')))

Jul  8 03:58:04 xcpng01 xapi: [20200708T08:58:04.897Z|audit||58846 |Async.host.power_on R:19f454d900b0|audit] ('trackid=68849bbe6e0a22d9b0b024c21c2b6e36' 'LOCAL_SUPERUSER' 'root' 'ALLOWED' 'ERROR:INTERNAL_ERROR: [ (Failure \"The host failed to power on.\") ]' 'API' 'host.power_on' (('host' 'xcpng02' '1139bfbc-c487-4311-8206-bbffe96ed6b1' 'OpaqueRef:426e950f-f926-490f-b598-328314aafbdc')))

Now going back in time to audit.log.[2-7].gz

olivierlambert

It's possible that XAPI recorder the host.power_on and loop on trying to do that, but to be fair, I never saw that before

Can you power on that host to see if it stops?

dsiminiuk

@livierlambert

Going back as far as I can to audit.log.7.gz only takes me back to June 30 and this has been a problem for more than a month so I don't think I can find the original request.

dsiminiuk

@livierlambert said in Power on function:

Can you power on that host to see if it stops?

I had the whole rack powered off yesterday, 2 - XCP-NG 8.1 servers (DL380 Gen9), 1 Freenas (DL380 Gen9) for iSCSI VDIs. It still came back.

olivierlambert

So now everything is back online but despite that you got the task?

dsiminiuk

@livierlambert said in Power on function:

So now everything is back online but despite that you got the task?

Yes, everything is up and working, just the task appears in XO and also on the master's tasks list as we see.

olivierlambert

Weird. Can you disconnect XO from this pool, restart toolstack on master, and see if the task is popping again? I think it will, I don't see why XO would do that.

dsiminiuk

@livierlambert said in Power on function:

Weird. Can you disconnect XO from this pool, restart toolstack on master, and see if the task is popping again? I think it will, I don't see why XO would do that.

I would remove the hosts from XO, restart toolstack, add hosts back to XO? Is this correct?

olivierlambert

Don't add host back in XO for few minutes, and check if xe task-list got the task appearing without any XO connection.

dsiminiuk

@livierlambert

Host #2 has been powered off during these checks. I will power it on so that all pool members are available and try xe task-cancel again. Then proceed with your suggestion. Standby.

dsiminiuk

@livierlambert

OK this is interesting (host1 is master). I powered on host2 and ... the task disappeared from xe task-list and XO Tasks.

I powered off host2 again and the task is back.

[12:38 xcpng01 log]# xe task-list params=all
uuid ( RO)                  : 221a86ff-ae3f-d10e-26bc-9a7b693f44f0
            name-label ( RO): Async.host.power_on
      name-description ( RO):
            subtask_of ( RO): <not in database>
              subtasks ( RO):
           resident-on ( RO): cdb42b94-7ec6-47b4-aaef-d4c52f6bd7b3
                status ( RO): pending
              progress ( RO): 0.000
                  type ( RO): <none/>
                result ( RO):
               created ( RO): 20200708T17:37:00Z
              finished ( RO): 19700101T00:00:00Z
            error_info ( RO):
    allowed_operations ( RO): Cancel
    current_operations ( RO):
          other-config (MRW):

dsiminiuk

I am powering up host2 again only to remove it from XO and follow your direction.

dsiminiuk

@livierlambert

Both servers are running in the pool.
Both removed from XO.
Toolstack restarted on both servers via XCP-NG Center at 12:50pm local time.
Server communications re-established automatically in XCP-NG Center at 12:52pm.
Pool master added and 2nd server added to XO at 12:54pm.
No tasks are listed on either server with xe task-list.
No tasks are listed in the XO UI.
Powered off server 2 from the server's console screen (not XCP-NG Center or XO being the point).

I have a task in the master on the CLI and XO again.

uuid ( RO)                : 2f820b63-f03a-1258-1443-2f946569a4fb
          name-label ( RO): Async.host.power_on
    name-description ( RO):
              status ( RO): pending
            progress ( RO): 0.000

The only mention of power_on in audit.log since the power up time...

Jul  8 12:48:15 xcpng01 xapi: [20200708T17:48:15.096Z|audit||3666 |Async.host.power_on R:9802e63aaa34|audit] ('trackid=5835f391b350762fbb9dc4eb46b25614' 'LOCAL_SUPERUSER' 'root' 'ALLOWED' 'OK' 'API' 'host.power_on' (('host' 'xcpng02' '1139bfbc-c487-4311-8206-bbffe96ed6b1' 'OpaqueRef:426e950f-f926-490f-b598-328314aafbdc')))

Jul  8 12:48:15 xcpng01 xapi: [20200708T17:48:15.183Z|audit||4625 INET :::80|task.destroy D:39825944bdf5|audit] ('trackid=5835f391b350762fbb9dc4eb46b25614' 'LOCAL_SUPERUSER' 'root' 'ALLOWED' 'OK' 'API' 'task.destroy' (('self' 'Async.host.power_on' '221a86ff-ae3f-d10e-26bc-9a7b693f44f0' 'OpaqueRef:9802e63a-aa34-4f8e-9876-94f0b3778324')))

Danp

@siminiuk You should only be adding the pool master in XO, not the slaves.

dsiminiuk

@anp

http://xo/#/settings/servers is only for the pool masters not pool members?

I don't see where it says to not add pool members other than the master
https://xen-orchestra.com/docs/manage_infrastructure.html#home-view

I am grateful for the help but it didn't solve this issue. The task is back.

olivierlambert

Quoting the doc:

Add a host
Just click on "Add server", enter the IP of a XenServer host (ideally the pool master if in a pool)

So yes, in a pool, only the master is needed to connect to all machines.

jedimarcus

FYI, I have a server supporting WoL which I configured and everything worked like a charm.

Only complaint in XOA: the Host stays yellow in Busy (shutdown) Mode but even in this mode, the Start button is available and works.

We configured Wake on LAN (WoL) in the BIOS and tested it with a stupid Windows Tool. WoL seems to be way more stable and proven than any other methods and the only downside I see is that it's limited to a Local Range (as it works with MAC Addresses) but I wonder how many pools don't fit in that schema.