Summary
This article provides suggestions for troubleshooting error return code 10049 (CM_ERR_EADDRNOTAVAIL).
Resolution
This error indicates that one of the DPX modules is attempting to bind to an IP address that is not available on the server.
If a server has multiple IPs assigned to it, such as in a clustered environment, this error becomes more common. It may also occur if a server's address is changed after DPX was installed.
If you experience this problem:
1. Ensure that the address specified in the SSICMAPI environment variable reflects a correct, permanent address for the server generating the error.
The IP address is specified with the argument -hn <IP address>, and is not specified in a default installation for non-clustered servers. Any changes to this environment variable require that DPX is stopped and restarted before the changes take effect.
Windows:
This is set in the registry in HKLM/Software/Syncsort/Agent/SSICMAPI.
Example: -a 6123 -sbs 65535 -winm -hn 192.168.1.50
The IP address specified should show up in ipconfig/all output, and must be a permanent address for this server.
UNIX:This is set in the text file $SSPRODIR/misc/SSICMAPI, where $SSPRODIR is the directory where DPX is installed.
Example: -a 6123 -sbs 65535 -hn 192.168.1.50
The IP address specified should show up in ifconfig <interface> output (e.g., ifconfig hme0 or, on some OSes, ifconfig -a to show all), and must be a permanent address for this server.
2. For a Windows cluster, also check that the environment variable SSICMAPI on each node in the cluster correctly reflects the cluster address. In these cases, this is not an optional variable.
Note that cluster support applies only to DPX version 2.1.5 or higher.
Windows:
The address is set in HKLM/Software/Syncsort/BackupExpress/<cluster_hostname>/SSICMAPI and it must be the IP address associated with the Backup Express clustered resource group.
3. Verify that name resolution on this server is reporting the correct permanent IP address for this machine. This is particularly important in cluster environments.If DNS is not available, update the hosts information to reflect correct information so that the server can resolve its own hostname, both the short form and the fully qualified name (for example, both myserver and myserver.example.com resolve to the correct address). Updates to the hosts files should be immediately accessible, but may be cached by software already running. For this reason, it is best to restart Backup Express after making any changes to the hosts information.
Example:
192.168.1.50 myserver.example.com # FQDN
192.168.1.50 myserver # short hostname
Windows:
This information is stored in the text file %%WINNT%%\System32\Drivers\ETC\HOSTS.
UNIX:This information is stored in the text file /etc/hosts.