[rancid] ERROR: sh: cisco: command not found

[Simone Felici]

Hello to all!

I'm having an issue with rancid. sometimes I get this errors in the logs:

sh: cisco: command not found
sh: cisco: command not found
sh: cisco: command not found
sh: cisco: command not found
(...)

Then the script doesn't process all the list of devices!
If I execute manually the last device rancid has processed as well as the device immediate after, I
have no errors. The list is long and it uses another server to collect these configurations via ssh
(usercmd, usercmd_chat). Could it be a timeout issue on the remote server used to contact the
device? Or what is happen? How can I debug?

The script ends in the log with:

Transmitting file data ..................................
Committed revision 21106.
+ rm -f /tmp/.customer_cpe.run.lock

Also seems it ends to process the list, but isn't true, because the list in routers.up is longer.

Thanks a lot!

Simon

[Simone Felici]

Hello,

Post by Simone Felici
Hello to all!

what version of rancid on what o/s?

rancid 2.3.6 on CentOS 5.8 (server and slaves)

Post by Simone Felici
sh: cisco: command not found
sh: cisco: command not found
sh: cisco: command not found
sh: cisco: command not found
(...)
Then the script doesn't process all the list of devices!
If I execute manually the last device rancid has processed as well as the device immediate after, I
have no errors. The list is long and it uses another server to collect these configurations via ssh

how long? does it exceed the command-line length allowed by your o/s?
perhaps a quoting error?

Are they not processed one per time?
the routers.all contains 672 entries, routers.up -> 572.

but, what you describe is not something we support, so you've presumably
modified something to do this, but not told us what or provided the code.

The installation is old. The only I've edited a while ago was add the "out of band" access via
"usercmd" "usercmd_chat" options. Is this availabe in the last release?
So I could update.

Post by Simone Felici
(usercmd, usercmd_chat). Could it be a timeout issue on the remote server used to contact the
device? Or what is happen? How can I debug?
Transmitting file data ..................................
Committed revision 21106.
+ rm -f /tmp/.customer_cpe.run.lock
Also seems it ends to process the list, but isn't true, because the list in routers.up is longer.

what makes you think that it has not processed the entire list?

Because I've some devices I've edited and the differences are not saved, even if they should be
processed. They are in the routers.all db. Only if I call the script for a single device it works:

bin/rancid-run -r <device_name> <repo_name>

Then, I've edited temporary rancid-fe to enable dedug:

(...)
} else {
exec($vendortable{$vendor} . " -d $router");
}
(...)

This logs me all the HIT commands as well the command that is executed to contact the device.
At a certain point I get:

HIT COMMAND:<my_device>#write term
In WriteTerm: <my_device>#write term
sh: cisco: command not found
sh: cisco: command not found
(...line repeated many times..)
=====================================
Getting missed routers: round 1.
(...)
sh: cisco: command not found
(...)
EOF


Thanks for the help,

Simon