[rancid] Setup issue

[Alan McKinnon]

Hi Michael,

By themselves those errors are not harmful (rancid can deal with them).
To show why, I need to backtrack a bit and explain how the code is
organised. This applies to 2.3.x I haven't read 3.0 enough to comment much.

A text dump of a login session running commands show the structure:
a text prompt, followed by the prompt character (> or #) then the
command run, a newline, followed by the command output. Repeat for all
commands. Finally, and exit, quit or similar command and the end of the
text.

The rancid script starts by skipping over everything from the beginning
till it finds the defined prompt character. It records that text as the
separator between the end of one command and the start of the next. The
command run is immediately after on the same line which it looks up in
@command and the value of that key is the sub to pass the text to. The
script passes all text it finds to the sub until it hits the next
prompt; and the cycle repeats.

Two things can go wrong here:
- A peculiar prompt that confuses the script[1]
- the sub is in a while or other loop when that command output ends,
without this the code that detects the prompt does not run when it must.
The rule is that the command subs must be running their top-level code
when the command exits.

My first guess is that the command before those that didn't run is not
exiting cleanly. Run this:

clogin -t 90 -c "show version;show memory;show diag"

and look for oddities in the output.

Ordinarily a failed command is not a problem. The script looks for error
messages and if it finds say "command not authorized" it simply skips
over it and continues with the next one.



[1]As an example I had a weird bug once with Cisco that didn;t run
classic IOS. A command "show admin "something-or-other" didn't have a
trailing newline in the output; all other commands were correct. My
solution is to comment that command out - for me the output isn't
especially useful in any way.



On 03/05/2014 13:45, Michael T. Voity wrote:
> Alan,
> 
> I ran the xrancid and was given this -
> 
> [rancid at netwatch bin]$ ./xrancid -d watt-uvm-gw3-200.3.uvm.edu
> executing clogin -t 90 -c "show version;show memory;show diag;show
> switch;show slot;show configuration detail;show configuration"
> watt-uvm-gw3-200.3.uvm.edu
> sh: clogin: command not found
> watt-uvm-gw3-200.3.uvm.edu: missed cmd(s): all commands
> watt-uvm-gw3-200.3.uvm.edu: End of run not found
> watt-uvm-gw3-200.3.uvm.edu: End of run not found
> 
> Then I ran just -
> 
> ./clogin -t 90 -c "show version;show memory;show diag;show switch;show
> slot;show configuration detail;show configuration"
> watt-uvm-gw3-200.3.uvm.edu
> 
> And determined that-  show switch;show slot;show configuration
> detail;show configuration did not run.
> 
> 
> Ideas?
> 
> -Mike
> 
> 
> Michael T. Voity
> Network Engineer
> University of Vermont
> 
> On 5/3/14, 2:43 AM, Alan McKinnon wrote:
>> That's the standard log message you posted; for debugging purposes it's
>> not especially useful. It's telling you that stuff didn't work and then
>> lists all the stuff that didn't work. It doesn't indicate why.
>>
>> Run the parser script directly in debug mode to find out what's really
>> going on.
>>
>> xrrancid -d <hostname>
>>
>>
>> This might not give you the real failure yet, but it will narrow it down
>> to which area the problem happens in.
>>
>>
>> On 02/05/2014 21:11, Michael T. Voity wrote:
>>> Hello,
>>>
>>> I'm new to the RANCID realm.   Before posting to this list I have done
>>> my research and cannot quite seem to get the right answer to my issue.
>>>
>>> I'm trying to get RANCID to work with my Cisco ASR9001 which has
>>> ios-xr -
>>>
>>> In the log file i get these messages -
>>>
>>> Trying to get all of the configs.
>>> swick-uvm-gw2: missed cmd(s): dir /all disk1a:,admin show diag,dir /all
>>> slavedisk2:,dir /all sec-slot2:,dir /all disk1:,dir /all sec-nvram:,dir
>>> /all disk2:,dir /all sec-bootflash:,dir /all slaveslot2:,dir /all
>>> disk0:,dir /all compactflasha:,dir /all slaveslot0:,dir /all
>>> sec-slot1:,dir /all harddiska:,dir /all slavenvram:,dir /all
>>> sec-disk2:,dir /all sec-disk0:,dir /all harddiskb:,dir /all
>>> slavedisk1:,show controllers,show rpl maximum,dir /all slavedisk0:,show
>>> debug,dir /all bootflash:,dir /all sec-slot0:,dir /all sec-disk1:,dir
>>> /all disk0a:,dir /all slot2:,dir /all compactflash:,dir /all
>>> harddisk:,dir /all slot0:,admin show running,show vlan,dir /all
>>> slavebootflash:,dir /all slaveslot1:,dir /all nvram:,admin show
>>> inventory raw,show running-config,dir /all slot1:
>>> swick-uvm-gw2: End of run not found
>>> !
>>> watt-uvm-gw3: missed cmd(s): dir /all disk1a:,admin show diag,dir /all
>>> slavedisk2:,dir /all sec-slot2:,dir /all disk1:,dir /all sec-nvram:,dir
>>> /all disk2:,dir /all sec-bootflash:,dir /all slaveslot2:,dir /all
>>> disk0:,dir /all compactflasha:,dir /all slaveslot0:,dir /all
>>> sec-slot1:,dir /all harddiska:,dir /all slavenvram:,dir /all
>>> sec-disk2:,dir /all sec-disk0:,dir /all harddiskb:,dir /all
>>> slavedisk1:,show controllers,show rpl maximum,dir /all slavedisk0:,show
>>> debug,dir /all bootflash:,dir /all sec-slot0:,dir /all sec-disk1:,dir
>>> /all disk0a:,dir /all slot2:,dir /all compactflash:,dir /all
>>> harddisk:,dir /all slot0:,admin show running,show vlan,dir /all
>>> slavebootflash:,dir /all slaveslot1:,dir /all nvram:,admin show
>>> inventory raw,show running-config,dir /all slot1:
>>> watt-uvm-gw3: End of run not found
>>> !
>>>
>>>
>>> When i switch it from ios-xr to ios in the router.db file, it works, but
>>> every time I run the rancid-run it gives me a huge diff of the ios-xr
>>> routers.
>>>
>>> I have also un-commented  NOPIPE=YES; export NOPIPE
>>>
>>> I have setup my ios-xr routers in a separate group.
>>>
>>> Still no dice.
>>>
>>> I have a separate group for my ios devices that is working fine.
>>>
>>> ideas?
>>>
>>> -Mike
>>>
>>
> 
> _______________________________________________
> Rancid-discuss mailing list
> Rancid-discuss at shrubbery.net
> http://www.shrubbery.net/mailman/listinfo/rancid-discuss
> 
> 


-- 
Alan McKinnon
alan.mckinnon at gmail.com