Recovery from RLINK Connect Problems

This section describes the errors that may be encountered when connecting RLINKs. To be able to troubleshoot RLINK connect problems, it is important to understand the RLINK connection process.

Connecting the Primary and Secondary RLINKs is a two-step operation. The first step, which attaches the RLINK, is performed by issuing the vxrlink att command or the vradmin startrep command. The second step, which connects the RLINKs, is performed by the kernels on the Primary and Secondary hosts.

The attach command performs a number of checks to ensure that the operation is likely to succeed, and if it does, the command changes the state of the RLINKs from DETACHED/STALE to ENABLED/ACTIVE. The command then returns success.

If the command is successful, the kernel on the Primary is notified that the RLINK is enabled and it begins to send messages to the Secondary requesting it to connect. Under normal circumstances, the Secondary receives this message and connects. The state of the RLINKs then changes from ENABLED/ACTIVE to CONNECT/ACTIVE.

If the RLINK does not change to the CONNECT/ACTIVE state within a short time, there is a problem preventing the connection. This section describes a number of possible causes. An error message indicating the problem may be displayed on the console.

• If the following error displays on the console:

VxVM VVR vxrlink INFO V-5-1-5298 Unable to establish connection with remote host <remote_host>, retrying

Make sure that the vradmind daemon is running on the Primary and the Secondary hosts; otherwise, start the vradmind daemon by issuing the following command:


  #  /sbin/init.d/vras-vradmind.sh start
• If the following error displays on the console:

VxVM VVR vxrlink ERROR V-5-1-3504 Primary datavol hr_dv01 is not mapped on secondary, yet

Make sure that a corresponding Secondary volume mapped to the Primary data volume hr_dv01 exists on the Secondary. If the volume has a different name, either rename the volume, or map the name of the volume on the Primary with the name of the volume on the Secondary. To map the volume on the Secondary with the volume on the Primary, issue the following command on the Secondary:


  # vxedit -g hrdg set primary_datavol=hr_dv01 hr_dv01_sec

where hr_dv01_sec is the name of the volume on the Secondary.

• If there is no self-explanatory error message, issue the following command on both the Primary and Secondary hosts:

  # vxprint -g diskgroup -l rlink_name

In the output, check the following:

The remote_host of each host is the same as local_host of the other host.

The remote_dg of each host is the same as the disk group of the RVG on the other host.

The remote_dg_dgid of each host is the same as the dgid (disk group ID) of the RVG on the other host as displayed in the output of the vxprint -l diskgroup command.

The remote_rlink of each host is the same as the name of the corresponding RLINK on the other host.

The remote_rlink_rid of each host is the same as the rid of the corresponding RLINK on the other host.

Make sure that the network is working as expected. Network problems might affect VVR, such as prevention of RLINKs from connecting or low performance. Possible problems could be high latency, low bandwidth, high collision counts, and excessive dropped packets.

• For an RLINK in private disk group, issue the following command on each host.

  # ping <remote_host> 

There should be no packet loss or very little packet loss. To ensure that the network can transmit large packets issue the following command on each host for an RLINK in a private disk group.


  #  ping remote_host 8192 

The packet loss should be about the same as for the earlier ping command.

• Issue the vxiod command on each host to ensure that there are active I/O daemons. If the output is 0 volume I/O daemons running, activate I/O daemons by issuing the following command:

        #    vxiod set 10
• VVR uses well-known ports to establish communications with other hosts.

Issue the following command to display the port number:


  # vxprint -g diskgroup -l rlink_name 

Issue the following command to ensure that the heartbeat port number in the output matches the port displayed by vxprint command:


  # vrport

Confirm that the heartbeat port has been opened by issuing the following command:


  # netstat -an | grep port-number 

where port-number is the port number being used by the heartbeat server as displayed by the vrport command.

The output looks similar to this:


     udp     0   0  *.port-number