When a clustered metadata server node encounters communication errors, it disconnects from the master node. When it attempts to rejoin the cluster, the node can run out of file handles.

If you encounter this issue, the following message is returned in the SASMetadataServer*.log files:

2015-04-21T22:02:41,591 ERROR [00000015] :sas The TCP/IP tcpSockAccept support routine failed with error 24 
(Too many files are open.)

and/or 

2015-04-05T16:23:14,479 INFO  [01331883] 63135:sas- INTERNAL ERROR: Connecting node SASMeta -
Metadata Server has no pIface.

To work around this issue, complete the following steps:

1. Ensure that the network is functioning and has no errors.
2. Stop and start the metadata server node that is offline. Use the following command to control the SAS Metadata Server:
   SAS-configuration-directory/Lev1/SASMeta/MetadataServer/MetadataServer.sh options
3. Use SAS^® Management Console to review the state of the cluster:
   
   • Expand the Metadata Manager folder.
   • Right-click Active Server Properties.
   • Click the Cluster tab.
     

     Note: If this tab is not visible, you do not have a metadata server cluster.

   • If everything is running well, the tab should display the following:
     

     State: QUORUM
     Defined nodes=3
     Current nodes=3

   • If any nodes are not running, review the SASMetadataServer*.log files and resolve the issue that is preventing them from running.

Click the Hot Fix tab in this note to access the hot fix for this issue.

Operating System and Release Information

A fix for this issue for Base SAS 9.4_M2 is available at:

A fix for this issue for Base SAS 9.4_M1 is available at: