Original Publishing Date:
2020-01-22

Symptoms

shaman stat command prints a warning:

shaman: Failed to read XML model: syntax error

What does it mean and how to fix this kind of error?

Cause

The error is not harmful and appears when there are "Unknown" resources:

~# shaman -c pcs stat
    Failed to read XML model: syntax error
     Cluster 'pcs'
     Nodes: 3
     Resources: 5

        NODE_IP           STATUS      RESOURCES
     *  10.1.1.10       Active      2 CT, 0 Unknown
      M 10.1.1.11       Active      1 CT, 0 Unknown
        10.1.1.12       Inactive    1 CT, 0 Unknown

          CT ID                    PWRR        STATUS      OWNER_IP          PRIORITY
             101                   on          Active      10.1.1.11         0
             102                   on          Active      10.1.1.13         0
             103                   on          Active      10.1.1.12         0
             104                   on          Active      10.1.1.11         0

     Unknown STATUS      OWNER_IP          PRIORITY
             Broken      Unknown           0

Resolution

Clean up "Unknown" and "Broken" resources:

# shaman cleanup-broken

This command attempts to relocate all broken resources to a local node without starting them. After relocation these resources can be manually migrated to desired nodes. Files of broken resources, which cannot be relocated, are deleted.

For other cases, to get rid of the error in shaman stat, the resources should be placed under appropriate node in shaman records.

1. Get "Node ID” (or host_id) value for each PCS host in cluster by running "shaman stat" with verbose output ("-v" option"):

   # shaman -v -c pcs stat
   Failed to read XML model: syntax error
    Cluster 'pcs'
    Nodes: 3
    Resources: 4
   
    NODE_IP           STATUS      NODE_ID            RESOURCES
   
    *M 10.1.1.37       Active      34705c9745f742a1   1 CT, 1 VM, 0 Unknown
       10.1.1.38       Active      da6fd9b1b4c34fb6   0 CT, 0 VM, 0 Unknown
       10.1.1.39       Inactive    42064ead9c244d74   1 CT, 0 VM, 1 Unknown
   
     CT ID         PWRR        STATUS      OWNER_IP          OWNER_ID         PRIORITY
        101        on          Active      10.1.1.39       42064ead9c244d74   0
        999        on          Active      10.1.1.37       34705c9745f742a1   0
   
     VM NAME       PWRR        STATUS      OWNER_IP          OWNER_ID         PRIORITY
        vm-test    off         Active      10.1.1.37       34705c9745f742a1   0
   
   
    Unknown STATUS      OWNER_IP          OWNER_ID           PRIORITY
            Active      10.1.1.39       42064ead9c244d74   0

2. The list of shaman resources is located under the /pstorage/$CLUSTER_NAME/.shaman/md.$NODE_ID/resources directory.

   Since the directory is under Pstorage mount, it can be accessed from any node of the cluster. Replace $CLUSTER_NAME and $NODE_ID in the example above with appropriate values.

3. For existing virtual machines, their resource files should be moved under appropriate PCS node:

   ~# mv /pstorage/$CLUSTER_NAME/.shaman/md.$HOST_ID_1/resources/vm-NAME /pstorage/$CLUSTER_NAME/.shaman/md.$HOST_ID_2/resources/vm-NAME
   

   Where md.$HOST_ID_1 is the host improperly owning this VM and md.$HOST_ID_2 is the proper host running this particular VM.

   Note: Do not change the name for the resource file, it is always written in vm-NAME or ct-CTID format.

Internal content

Link on internal Article