CiscoWorks LMS 3.2 SCP failure

netzeatkw · ‎06-08-2010

Hello,

I changed the transport protocol for images from tftp to scp only. In Common Services, Server Admin I have defined a scp user and password.

The ip scp server is enabled on the network components.

Now scp works well from the command line of the switch in both direction to and from the CiscoWorks Server.

In RME Software management, Software Repository, the download of Images via scp works well.

The upload in RME Software distribution fails, here is the log:

JOB ID - 1727 SUMMARY
---------------------------------------------

Device Name : c356005xxzz
Software upgrade result - Failed

Total number of devices scheduled for upgrade : 1
Number of devices upgraded : 0
Number of devices failed : 1
Total time taken for Job - 7 Minutes 20 Seconds

Status of Job - FAILED

-----------------------------------------------------------------------------------------------

Starting Job: ID - 1727
Job Started At : Tue Jun 08 14:45:32 CEST 2010

Job running on : SIZZ0000.zzintfeld.net

Upgrade Mode : Parallel
Reboot Mode : Parallel
Protocol Order for Image Transfer : SCP
Protocol Order for Config-operations : SSH,SCP

Total number of device(s) to be processed : 1

--------------------------------------------------------------------------------
Processing Device : c356005xxzz
Start Time : Tue Jun 08 14:46:42 CEST 2010

Device is locked for exclusive access.

The supported protocols for image transfer are: SCP

Connecting to the device using SSH...
Connected successfully.

Sufficient free space is available on flash to copy the image.

Retrieving configuration file from the device...
Current device config is copied to D:\Apps\CSCOpx\files\rme\jobs\swim\1727\c356005n0zz_Config

Checking if HTTP server is enabled on the device...
Http Server is not enabled, will continue with Software Upgrade Operation.

Copying c3560-ipbasek9-mz.122-46.SE.bin from Software Repository to scp-specific directory.
Copied successfully to D:/Apps/CSCOpx/temp/rep_sw_5918160919176062293

Loading image file to flash device : rep_sw_5918160919176062293 --> flash:c3560-ipbasek9-mz.122-46.SE.bin using SCP
Error loading image to Flash. See D:\Apps\CSCOpx\files\rme\jobs\swim\1727\c356005n0zz_telnet.log for the output from the device on which the image transfer failed.
Image Copy Operation Failed

Device is unlocked.

Device Upgrade Result : Failed
End Time:Tue Jun 08 14:50:43 CEST 2010

--------------------------------------------------------------------------------

In the syslog of the switch, the following message appears:

.Jun 8 14:51:15.676: SSH2 1: alloc 37888 not supported

The scp image transfer only fails in direction from the cisco works repository to the switch.

A "copy scp: flash:" with the scp user from the switch command line works well, the file is copied from the home directory of cisco works.

In the log above, RME copies the file temporarily in a temp directory under this home directory.

Any idea how this could be solved?

Thanks in advance.

Wolfgang

Joe Clarke · ‎06-09-2010

Post the D:\Apps\CSCOpx\files\rme\jobs\swim\1727\c356005n0zz_telnet.log mentioned in the error output as well as the swim_debug.log from the same directory.

netzeatkw · ‎06-10-2010

This are the requested logs.

Joe Clarke · ‎06-10-2010

In this case, the device is the SCP server. Try testing an SCP copy from the server to the device. See if that completes successfully. You might also try enabling "debug ip scp", then running the SWIM job to see what debug output is shown when the copy dies.

netzeatkw · ‎06-11-2010

A scp session from the ciscoworks server hangs after successful authentication:

winscp> open scp://c356005n0zz

Searching for host...

Connecting to host...

Authenticating...

Username: username

Using keyboard-interactive authentication.

Password:

Authenticated.

Starting the session...

Host is not communicating for more than 15 seconds. Still waiting...

Warning: Aborting this operation will close connection!

(A)bort: Abort

Terminated by user.

The logging on the switch shows the following:

*Mar 2 18:34:19.664: SCP: [22 -> 172.22.11.111:22880] send

*Mar 2 18:34:19.672: SCP: [22 <- 172.22.11.111:22880] recv C0664 8811199 rep_sw_6035175346357903795

*Mar 2 18:34:19.672: SCP: [22 -> 172.22.11.111:22880] send

*Mar 2 18:34:34.864: SSH2 1: alloc 37888 not supported

*Mar 2 18:34:34.873: SSH2 1: alloc 37888 not supported

Joe Clarke · ‎06-11-2010

What version of code is this switch running now?

netzeatkw · ‎06-14-2010

c3560-ipbasek9-mz.122-53.SE.bin

Joe Clarke · ‎06-14-2010

To be clear, you're trying to downgrade the switch using SWIM?

netzeatkw · ‎06-14-2010

Yes, it is in a testbed where I want to downgrade to our productive version (IOS 12.2(46)SE).

Joe Clarke · ‎06-14-2010

I think you might have found a new bug. I recommend you open a TAC service request so that this can get reproduced.

mikik · ‎03-01-2011

Hi:

We are running into the exact same problem with Cisco 4500 ... SCP starts working and the receive Alloc 37888 not supported.

Has a solution been found? Was a Cisco Tac case opened? I searched the Cisco Bug data base and supprizeingly there are no SCP related bugs with LMS 3.2 . The bug search seems to me suffering from some bug of its own.

Thanks for any help

mickey

Joe Clarke · ‎03-01-2011

I do not think a service request was opened. I also don't see any bugs for LMS 3.2 (or any device-specific bugs) that have been filed for this. I think it would be beneficial to have this analyzed by TAC.