In a SAS Metadata Server cluster, you might encounter a situation where a slave node is reported to be a part of the quorum, but it does not receive redirection requests from the master node.

This situation is most likely to occur when Enterprise Grid Orchestrator (EGO) is used to start the metadata server cluster nodes or if the metadata servers on each node are all started at approximately the same time. This problem occurs on start-up of the metadata servers when they are trying to connect to each other in order to determine which server will become the master and which server will be the slave.

The issue might also occur when a quorum loss occurs during normal run time. After the quorum loss, if two cluster members both connect to a third cluster member at almost the same time, this situation might occur where one slave joins the quorum without receiving future redirection requests from the master node.

In either situation, the following messages will be seen in the metadata server log of one of the nodes in the metadata server cluster:

This problem might be hard to detect at run time, because it requires a review of the metadata server logs. To determine whether you are encountering this problem, you must check the metadata server master node log to see whether redirect requests are going to all slaves or to only a subset of slave nodes.

When this problem occurs, the following message is logged to the metadata server master node log file (SASMeta_MetadataServer_date_host_pid.log) when the connection request is redirected to one of the slave nodes. In this case, the connection request is redirected to server "SASMeta - Metadata Server Node 3":

Under normal operating conditions, you should also see redirect requests go to the other slave node. Based on the log message above, the other slave node that receives redirection requests might be "SASMeta - Metadata Server Node 2" or "SASMeta - Metadata Server" depending on who the master is.

If you see that ALL of the redirection requests are going to only ONE slave node and not to the other, it is possible that you are encountering the problem described in this note.

To work around this problem, allow more time in between start-up of metadata server cluster nodes. This extra time prevents the condition where two nodes could encounter the error during start-up.

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_M3 is available at:

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