Issue with one of the cluster HB

[pradeep_r5]

Hi,

I have a two node cluster. HB is in Jeopardy state.

GAB Port Memberships
===============================================================
Port a gen   279c28 membership 01
Port a gen   279c28   jeopardy ;1
Port h gen   279c3f membership 01
Port h gen   279c3f   jeopardy ;1
 

One node show all linkare  up but second node is showing one link down

Unforunately one Node111, local-mac-address =false. We will set this true later.

Can somebody tell me Why one link down from one node only. If any issue, It should be down to both the node.

root@sun1 # lltstat -nvv
LLT node information:
    Node                 State    Link  Status  Address
   * 0 sun111           OPEN
                                  ce0   UP      00:14:4F:23:C3:66
                                  ce2   UP      00:14:4F:23:C3:66
     1 sun222          OPEN
                                  ce0   DOWN
                                  ce2   UP      00:14:4F:1E:88:55

root@sun222 # lltstat -nvv
LLT node information:
    Node                 State    Link  Status  Address
     0 sun111           OPEN
                                  ce0   UP      00:14:4F:23:C3:66
                                  ce2   UP      00:14:4F:23:C3:66
   * 1sun222           OPEN
                                  ce0   UP      00:14:4F:23:4F:71
                                  ce2   UP      00:14:4F:1E:88:55

 

[Marianne]

"We will set this true later."   Why only later? You can use eeprom command to change this.

Please provide output of 'dladm show-dev' on both nodes.

[pradeep_r5]

Hi Berg,

Actually this is a Production server and to activate this setting we required reboot . We are discussing this with management to schedule reboot of node.

These nodes are Solaris 9.

The Network Speed is 100 Mbps Full Duplex on both the node.

Regards

Pradeep Rajput

[Marianne]

The local-mac-address?=true is probably the cause of the incorrect link display.

Has this ever functioned correctly?

Have you tried to plumb IP addresses on the interfaces and test that way?

You can also test private network comms with "dlpiping" : http://www.symantec.com/docs/TECH19998

[Anoop_Kumar1]

Hello Pradeep,

Go for advise by Marianne. However a restart of llt services can also resolve the issue. I have face this issue sometimes and below resolved the issue most of times.

1. Freeze SG

2. Stop cluster on all nodes or sinle node where is problem

# hastop -force

3. Stop gab

# gabconfig -U

4. Stop LLT

# lltconfig -U

5. Unload gab and llt modules

# modunload -i <module>

6.  Start llt, gab and had on both server one by one in order

[Herve_GREGOIRE]

Hello,

I think the issue comes from local_mac_address :

• sun111 LLT driver is seeing 2 mac address, then 2 nic
• sun222 LLT driver is seeing a unique mac address, then 1 nic

Change your local_mac_address to true, and it will work.

Regards,

Herve