Having a Hard Time Getting Scheduled Nightly Config Backup to Complete Successfully

jasadelll

I've been wrestling with the getting the nightly config backup scheduled job to complete successfully.  When I check the logs, the first 20 or 30 or so fail, ususually with "Cannot download Running config : ERROR: Unable to Log into Router : Timeout. Device IP x.x.x.x"  or "ERROR: Cannot download Running config : Connection Refused by x.x.x.x".

The strange thing is that I can test SNMP and test Login for the so-called failed hosts and it works every time, and I can manually download startup and/or running config every time.  The failed hosts in the log are never the same ones.  They are always different.

I started out using Transfer Protocols (SSH auto/SSH auto/SCP) (for Execute Commands anc Scripts using/Request Configs using/Transfer Configs using).  Then I tried (SSH auto/SNMP/TFTP).  Then I tried (SSH auto/SSH auto/SSH auto), and finally (SSH1,SSH1,SSH1) but they all have the same results.

It almost looks like the job takes a few minutes (or seconds) to "ramp up" before it finally is able to successfully download the configs.  I'm really at a loss.  Anyone else having this problem?

rschroeder

I experienced similar issues; the best things I did to resolve the issue:

* Change the Settings/Configs/Config Transfers to 6 Simultaneous Downloads/Uploads

* Adjust the Device Inventory to only Inventory up to 10 devices concurrently

* Disable Real-Time Config Change Notifications

* Uncheck the Device Inventory settings that collected ARP and Bridge Table info

* Uncheck the Node Monitoring (up/down status and rediscovery)

*  Ensure the Settings/Network/Telnet&SSH settings have reasonable connection Timeout periods

* Turn on Session Tracing

* Break up the backup scheduled tasks into multiple jobs that can complete before the next scheduled job begins. 

It took a little trial & error to find the sweet spots, and reviewing the Session Traces helped, but it wasn't 100% intuitive.   But after making the changes I listed above, I no longer have any problems with scheduled backing up any of our switches, routers, load balancers, AP's, etc. 

Test backups at different times & break up the jobs into similar tasks.  Keep in mind that other network traffic may be causing your issues--for example, if your NCM server is being backed up to SAN or tape during the time you've scheduled an NCM job, you may have unexpected results.  Similarly, if you have any links with limited bandwith, and devices are being backed up to SAN or tape across those links when you're running NCM scheduled jobs, it could be something you'll have to schedule around.

Good luck!