I have the same problem where attempts to change configurations on any useful number of devices (50 should not be too many) will stop with "not attempted" on about half of them. For example, a job might stop at 30 not attempted and 27 "successful". Those ratios change slightly from job to job. The job is set to ignore failure and run in parallel. All of the devices pass all of the credentials tests in RME. The first few devices are successful with both configuration changes and saves. Then the devices begin to fail to save but all of the devices that are attempted are successful. This indicates that credentials are not a problem. These devices all work in smaller jobs. The problem seems to be that as the job continues in parallel, there is some sort of threshold that restricts telnet sessions. At some point the job simply gives up. Ciscoworks should handle thousands of devices not 10. The older version (2.2 command line) had similar save problems but didn't give up on jobs. I found a bug ID that was closed without resolution related to the save issue. It was reported in version 3.0 and I am running version 3.3. Basically there must be some kind of operating system restriction. The problem is not with credentials, it is not with network paths, it is not with AAA authentication, it is not with the number of TTY's available on the devices. I have ruled all of these out with logging, debugging, etc. Unfortunately, this issue has been somehow ignored and never published directly on Cisco's site.
I am running Sun Solaris 8 and RME 3.3. Anyone having similar issues? (I wrote a perl script that successfully telnets, configs, and saves to hundreds of Cisco devices so 10 is an unnacceptable number for a 5 digit priced software package).
