SeedDB crash when updating IP device whose serial has changed to a duplicate

[Jon Skarpeteig]

Attempting to change device type does not work.

Using version 3.6.0 installed by following debian install guide.

[Jon Skarpeteig (jon-skarpeteig)]

* Same behaviour in 3.6.3

[Morten Brekkevold (mbrekkevold)]

Unable to reproduce this on NAV 3.6.3. Please provide a full traceback of the error, if we're to have any chance of finding out what's going on.

[Jon Skarpeteig (jon-skarpeteig)]

MOD_PYTHON ERROR

ProcessId: 2336
Interpreter: 'nav'

ServerName: 'nav'
DocumentRoot: '/usr/share/nav/htdocs'

URI: '/seeddb/netbox/edit/predefined'
Location: None
Directory: '/usr/share/nav/htdocs/seeddb/'
Filename: '/usr/share/nav/htdocs/seeddb/netbox'
PathInfo: '/edit/predefined'

Phase: 'PythonHandler'
Handler: 'seeddb'

Traceback (most recent call last):

  File "/usr/lib/python2.5/site-packages/mod_python/importer.py", line 1537, in HandlerDispatch
    default=default_handler, arg=req, silent=hlist.silent)

  File "/usr/lib/python2.5/site-packages/mod_python/importer.py", line 1229, in _process_target
    result = _execute_target(config, req, object, arg)

  File "/usr/lib/python2.5/site-packages/mod_python/importer.py", line 1128, in _execute_target
    result = object(arg)

  File "/usr/share/nav/htdocs/seeddb/seeddb.py", line 165, in handler
    output = editPage(req,pageList[table](),request,unicodeError)

  File "/usr/share/nav/htdocs/seeddb/seeddb.py", line 456, in editPage
    selected)

  File "/usr/share/nav/htdocs/seeddb/seeddb.py", line 2672, in update
    box = device.netbox_set.all()

AttributeError: 'QuerySet' object has no attribute 'netbox_set'

MODULE CACHE DETAILS

Accessed: Tue Nov 9 09:27:41 2010
Generation: 0

_mp_fee3cb79bb902d1ae2e8fe8693adcb2a {
  FileName: '/usr/share/nav/htdocs/seeddb/initBox.py'
  Instance: 1 [IMPORT]
  Generation: 1
  Modified: Thu Nov 4 17:23:53 2010
  Imported: Tue Nov 9 09:27:41 2010
}

_mp_79759e91aa07f55e2ec9feb06dcdae38 {
  FileName: '/usr/share/nav/htdocs/seeddb/options.py'
  Instance: 1 [IMPORT]
  Generation: 2
  Modified: Thu Nov 4 17:23:53 2010
  Imported: Tue Nov 9 09:27:41 2010
}

_mp_ee36530af65736d4598346489c0aee92 {
  FileName: '/usr/share/nav/htdocs/seeddb/seeddb.py'
  Instance: 1 [IMPORT]
  Generation: 4
  Modified: Thu Nov 4 17:23:53 2010
  Imported: Tue Nov 9 09:27:41 2010
  Children: '/usr/share/nav/htdocs/seeddb/initBox.py',
                '/usr/share/nav/htdocs/seeddb/options.py',
                '/usr/share/nav/htdocs/seeddb/seeddbSQL.py'
}

_mp_6075681fed6c7055dc97214a475ff87e {
  FileName: '/usr/share/nav/htdocs/seeddb/seeddbSQL.py'
  Instance: 1 [IMPORT]
  Generation: 3
  Modified: Thu Nov 4 17:23:53 2010
  Imported: Tue Nov 9 09:27:41 2010
}

[Morten Brekkevold (mbrekkevold)]

That helped a lot more, thanks :)

This happens when the serial number that was found is already is associated with another IP Device you are monitoring. You're supposed to get a real error message about this, but a bug causes a crash instead.

[Morten Brekkevold (mbrekkevold)]

fix here: http://metanav.uninett.no/hg/series/3.6.x/rev/3e3fe909c11f

[Jon Skarpeteig (jon-skarpeteig)]

I noticed that the 'serial' field is filled with hostname when I attempt to update entry (after snmp reply). When I added entry, the serial got stored correctly. I am assuming this is related?

Behaviour on update (I dont' change anything, just hit update):
BayStack470-48T-PWR = Appears to be correct serial - but fails with the stacktrace above (but two others of the same type works)
catalyst37xxStack = Puts hostname in serial field - fails with stacktrace above
catWsCbs3032Del = Puts hostname in serial field - fails with stacktrace above
catalyst2950t24 = Puts hostname in serial field - fails with stacktrace above
However, different switches, of the type catalyst2950t24 works (probably configuration differences on switches, but unknown which)

Adding have worked, with two exceptions: one catalyst2950t24, and one catalyst37xxStack got added with hostname instead of serial - but for these, update works (with hostname detected as serial for update aswell)

None of the above have the same hostname or serial.. so I'm not sure that duplicate entries are the problem here?

[Morten Brekkevold (mbrekkevold)]

Are you saying that your devices' hostnames are being automatically filled in the serial number box when you update? I am unable to reproduce that on any version of NAV.

I do however see that if SeedDB is unable to retrieve a serial during an update, it will claim that it retrieved the box' existing serial number via SNMP (even though this isn't true). If it can't retrieve a serial number the first time you add a device, you are asked to provide it yourself. Some people type in the hostname as the serial in this case, you're sure you didn't do that?

[Jon Skarpeteig (jon-skarpeteig)]

I am sure. I have never been able to provide serial myself, as this field has been greyed out for all add and update actions I have performed. I also noticed a strange case where I deleted a valid entry I was unable to update, and added it again. On my new add it retrieved the hostname as serial, while it retrieved correct serial on the first time around.