Description:
This document lists a number of possible causes for the following 'windows socket' errors:
autoping:
AutoPinging Machine [hostname] AND checking the Remote Agent's DB Access.
read failed stream socket: Connection reset by peer
ERROR: AutoPing WAS NOT SUCCESSFUL!
chk_auto_up:
Checking Machine: hostname
read failed Win socket errno=10054
Could Not inspect machine: hostname
Solution:
The following are possible causes for windows socket errors:
- The Windows machine was not rebooted after installing the client as suggested in the AutoSys Windows Installation manual. Re-cycling the system will cause the new system Registry updates from the RA install to be read in by the OS.
- The auto_remote process that is being spawned by the Remote Agent service is dying due to system resource issues. A system reboot may also correct this condition as well.
- The auto_remote process is dying due to some other execution errors. A good troubleshooting method for this is to login to the Windows client and run 'auto_remote -x 'from the AutoSys Command prompt. This should return a version string. If it returns an error, then the same error is likely occurring when the auto_remote process is being launched by the RA service resulting in the Win Socket error.
For example, the result of "auto_remote -x" may be that it reports an error stating ntwdblib.dll was missing. This would indicate the need to check for the file's existence or location and copy the file from a different working Windows remote agent if needed.
- If "read socket failed error 10054" is received and the NT event viewer application log has an error"rc is -1 and Error is 10038":
Error 10038 means its an Invalid socket. This error usually occurs if any third party service providers are installed over TCP/IP. This can be checked with a utility called sporder.exe. If TCP/IP is not on top of the stack this error is generated, move TCP/IP to the top using sporder.exe and reboot the machine. autoping should be successful after these steps.
- The %PATH% variable is more than 1023 characters in length.
- If using an Oracle database, it may that the 'oci.dll' file is missing or corrupt. Copy the 'oci.dll' file from %ORACLE_HOME% to %AUTOSYS%\bin and restart the Remote Agent.
- The Remote Agent port may be configured differently between the Event Processor and Remote Agent machine. Make sure they are the same (5280 is the default).
- Verify that the port is available and listening. Command 'netstat -a' will help determine this. Also try to telnet to the port. There may be a firewall blocking the port, make sure it is open and able to communicate bi-directionally.