cancel
Showing results for 
Search instead for 
Did you mean: 

hastatus show state_admin wait

Home_224
Level 6

Dear All,

I have to manual modify the main.cf after hastop, to prevent Oracle service and listener startup by VCS, so I run the below step 

1. hastop 

2. Edit main.cf  and add "// " 

// requires group Oracle_MultiNIC online local soft
// Oraclde_sit_bkIP requires Oracle_sit_bkIP_NIC
// Oracle_NFS_DataFiles requires Oracle_sit_db_Mount
// Oracle_sit_app_Mount requires Oracle_sitdg_Diskgroup
// Oracle_sit_db_Mount requires Oracle_sitdg_Diskgroup

3. hacf -verify .

4. hastatus -sum

root@devuardbs01 # hastatus
attempting to connect....connected

group resource system message
--------------- -------------------- -------------------- --------------------
devuardbs01 STALE ADMIN WAIT: all systems stale


Can you advice what is the situation 

1 ACCEPTED SOLUTION

Accepted Solutions

CliffordB
Level 4
Employee
Hi.

Editing the main.cf should never be done directly. The file on disk is only read during initial cluster formation. The configuration is actually in memory.

Instead, either use the javaGUI or use the ha command set. Those commands take care of updating both the in-memory version and the version on dish on all nodes in the cluster. The Admin Guide provides details.

To “fix” your current situation, make sure you stop the cluster on all nodes. Check each node to make sure had and hadshadow processes are stopped. Verify your changes are still reflected in the main.cf on all nodes. It is likely that the changes did not take.

Restart the cluster. Assuming it comes up clean, follow the instructions in the admin guide to make changes with the ha commands. I think hares command will unlink resources. If you aren’t familiar with command line the javaGUI can be downloaded and installed on a Windows machine anywhere in your environment (works on laptops also) that has network access to the cluster.

_______________________________
My customers spend the weekends at home with family, not in the datacenter.

View solution in original post

3 REPLIES 3

CliffordB
Level 4
Employee
Hi.

Editing the main.cf should never be done directly. The file on disk is only read during initial cluster formation. The configuration is actually in memory.

Instead, either use the javaGUI or use the ha command set. Those commands take care of updating both the in-memory version and the version on dish on all nodes in the cluster. The Admin Guide provides details.

To “fix” your current situation, make sure you stop the cluster on all nodes. Check each node to make sure had and hadshadow processes are stopped. Verify your changes are still reflected in the main.cf on all nodes. It is likely that the changes did not take.

Restart the cluster. Assuming it comes up clean, follow the instructions in the admin guide to make changes with the ha commands. I think hares command will unlink resources. If you aren’t familiar with command line the javaGUI can be downloaded and installed on a Windows machine anywhere in your environment (works on laptops also) that has network access to the cluster.

_______________________________
My customers spend the weekends at home with family, not in the datacenter.

See:

https://www.veritas.com/support/en_US/article.100014008

So, after you verify the cluster is down on all nodes, make sure the main.cf is the same on all nodes.

Since you edited on just one node, that may have caused the ADMIN_WAIT state.

_______________________________
My customers spend the weekends at home with family, not in the datacenter.

Hi CliffordB,

 

Thank you for your reply!

Sorry I am not mention that the current OS is running in Solaris 9 . I using the command to do so,  I want to unlink the parent and child resources but I don't know how to identify the parent and child resource. 

Before I make the main.cf, using the haconf -makerw then go to edit main.cf  file . and check hacf -verify . to make sure there is no file error