cancel
Showing results for 
Search instead for 
Did you mean: 

Group vxfen keeps offline in VCS 5.1

joagmv
Level 4

 

Hi all!
 
We have started to setting io fencing up in our VCS in customized mode (using 3 CP servers). In VCS 6.0 everything works just fine but in VCS 5.1 (SP1RP2) the group vxfen keeps offline although it seems to be working properly:
 
Under Linux SLES 11:
 
 B  ClusterNOS      NODE1            Y          N               ONLINE
 B  ClusterNOS      NODE2            Y          N               OFFLINE
 B  vxfen           NODE1            Y          N               OFFLINE
 B  vxfen           NODE2            Y          N               OFFLINE
 
In the CPS, I can see both servers registered:
 
 CPS01TOR:/etc/VRTScps # cpsadm -s localhost -a list_nodes -c NOSTRUM
 Local node is CP Server, assuming nodeid as 0
 
 ClusterName      UUID                               Hostname(Node ID) Registered
 ===========   ===================================   ================  ===========
 NOS           {6176c10c-1dd2-11b2-8d00-a3e0b94e27bc}  NODE1(0)         1
 NOS           {6176c10c-1dd2-11b2-8d00-a3e0b94e27bc}  NODE2(1)         1
 
Log vxfend_A.log shows the proper registrations:
 
 End: join_local_node.sh returning SUCCESS (110)
 
No errors in any logs unless followings warnings:
 
 2013/01/26 12:44:03 VCS NOTICE V-16-1-10435 Group vxfen will not start automatically on System NODE2 as the system is not a part of AutoStartList attribute of the group.
 
The vxfen group only have a resource called coordpoint, which is online in both servers:
 
 NODE1:/var/VRTSvcs/log # hares -state coordpoint
 #Resource    Attribute             System     Value
 coordpoint   State                 NODE1      ONLINE
 coordpoint   State                 NODE2      ONLINE
 
The config we have is exactly the same that in other VCS (6.0) where the group vxfen is online.
 
Any idea about why the group vxfen keeps offline while it seems to be working properly?
 
Regards.
1 ACCEPTED SOLUTION

Accepted Solutions

mikebounds
Level 6
Partner Accredited

Amit is right in that CoordPoint type is of type None (persistent resource).  Normally for a persistent resource you put a phantom resource in the group so that you have a resource than can be online or offline so the group can be online and without this the group cannot be online.  However I have looked at the examples in 5.1 and 6.0 bundled agents guide and neither example has a phantom resource in the group.

So is you added a phantom resource to the group, then you would be able to online the group (and if you added AutoStartList group attribute, then it will auto online on boot), but I am perplexed as to why the examples do not include a phantom resource, especially as the manual does reflect they are aware of the issue - see extract from CoordPoint agent in bundled agents guide

 

AutoStartList attribute
AutoStartList is a service group attribute that needs to be populated with a
system list. The VCS engine brings up the specified service group on the nodes
in the list.
AutoStartList is not a required attribute for the service group that contains the
CoordPoint resource. The CoordPoint resource is a persistent resource and when
a service group is configured with this type of resource, it cannot be brought
online.
Specifying the AutoStartList with a system list does not change the behavior of
the service group. The service group will be reflected in OFFLINE status itself,
irrespective of the AutoStartList attribute.
 

 

So this is saying the group is expected to be OFFLINE, so I don't know why it shows as online in your 6.0 cluster, which I think is what you are saying.

Mike

View solution in original post

8 REPLIES 8

mikebounds
Level 6
Partner Accredited

Have you tried to online the vxfen group (hagrp -online vxfen -any) and it fails, or are you just saying it is not starting automatically.  Could you most extract from /etc/VRTSvcs/conf/config/main.cf to showing vxfen group - it be just that AutoStart or AutoStartList is not set correctly.

Mike

joagmv
Level 4

Yeah, I tried almost everything. The config is:

 

group vxfen (
        SystemList = { NODE1= 0, NODE2= 1 }
        AutoFailOver = 0
        Parallel = 1
        )
 
        CoordPoint coordpoint (
                )
 
        // resource dependency tree
        //
        //      group vxfen
        //      {
        //      CoordPoint coordpoint
        //      }
 
Same config that in others cluster, which is the default config that the installation script creates.
 
Manually...
 
 NODE1:~ # hagrp -online vxfen -sys NODE1
 VCS NOTICE V-16-1-50036 There are no enabled resources in the group vxfen to online
 
As posted before, resource coordpoint is online and working properly.
 
I tried to define all the typical parameters as SystemList, AutoStartList but nothing changed (although the NOTICE was different).
 
I was checking even the fixed bugs in next versions just in case this is a bug but nothing found.
 
this is the default installation... and it is happening in two clusters with VCS 5.1XXXX.

 

arangari
Level 5

i think CoordPoint type is of type None. Try 'hatype -dispaly CoordPoint | grep Operations'.

 

 

joagmv
Level 4

It is of type None:

 

  NODE1:~ # hatype -display CoordPoint | grep Operations
  CoordPoint   Operations             None
 
but in a VCS 6.0 where vxfen is ONLINE, the resource CoordPoint is type None too. I compared all the parameters and are identical (well, technically, 6.0 have 2 or 3 new parameters.... :) )
 
 

mikebounds
Level 6
Partner Accredited

Amit is right in that CoordPoint type is of type None (persistent resource).  Normally for a persistent resource you put a phantom resource in the group so that you have a resource than can be online or offline so the group can be online and without this the group cannot be online.  However I have looked at the examples in 5.1 and 6.0 bundled agents guide and neither example has a phantom resource in the group.

So is you added a phantom resource to the group, then you would be able to online the group (and if you added AutoStartList group attribute, then it will auto online on boot), but I am perplexed as to why the examples do not include a phantom resource, especially as the manual does reflect they are aware of the issue - see extract from CoordPoint agent in bundled agents guide

 

AutoStartList attribute
AutoStartList is a service group attribute that needs to be populated with a
system list. The VCS engine brings up the specified service group on the nodes
in the list.
AutoStartList is not a required attribute for the service group that contains the
CoordPoint resource. The CoordPoint resource is a persistent resource and when
a service group is configured with this type of resource, it cannot be brought
online.
Specifying the AutoStartList with a system list does not change the behavior of
the service group. The service group will be reflected in OFFLINE status itself,
irrespective of the AutoStartList attribute.
 

 

So this is saying the group is expected to be OFFLINE, so I don't know why it shows as online in your 6.0 cluster, which I think is what you are saying.

Mike

mikebounds
Level 6
Partner Accredited

Really, a service group is not ONLINE or OFFLINE if there are no non-persistent resources in the group, so I guess it could show ONLINE as well as OFFLINE, but it does normally show as OFFLINE.  So you could either ignore the OFFLINE state (and only worry if resource is in a faulted state) or add a phantom resource (which you may want to check with Symantec first as vxfen group may be classed as a special group which you shouldn't change).

Mike

joagmv
Level 4

It makes sense because the resource Phantom is defined in the clusters which have vxfen ONLINE :)

I did not notice about this agent in the config. It is probable to configure more clusters this weekend, so I will post the results.

Thanks Amit and Mike, today I learned a new feature of VCS :)

 

 

joagmv
Level 4

Tested and worked!!! Defining the resource phantom resolved the "problem".