[hobbit] Re: Devmon causing core dumps

[Buchan Milne]

On Monday 03 November 2008 09:46:42 Buchan Milne wrote:
> On Friday 31 October 2008 19:55:35 Robert Holden wrote:
> > When I run snmpwalk on my interfaces I get the following:
> > (I believe that the devmon template is using this ... notice that they
> > are all the same.
>
> Depending on the SNMP implementation (and may differ on different IOS
> versions).
>
> > A transform won't help)
> > IF-MIB::ifName.70 = STRING: AT5/0/0
> > IF-MIB::ifName.71 = STRING: AT5/0/0
> > IF-MIB::ifName.72 = STRING: AT5/0/0
> > IF-MIB::ifName.73 = STRING: AT5/0/0
> > IF-MIB::ifName.74 = STRING: AT5/0/0
>
> One of our 7613's has:
>
> IF-MIB::ifName.1 = STRING: Gi3/1
> [...]
> IF-MIB::ifName.128 = STRING: ATM10/1/0.0-atm subif
> IF-MIB::ifName.129 = STRING: ATM10/1/0-aal5 layer
> IF-MIB::ifName.130 = STRING: ATM10/1/0.0-aal5 layer
> IF-MIB::ifName.131 = STRING: VLAN-1
>
> > (This could be used, but you would end up w/ very long names.  G0/0 would
> > become GigabitEthernet0/0)
> > IF-MIB::ifDescr.70 = STRING: ATM5/0/0
> > IF-MIB::ifDescr.71 = STRING: ATM5/0/0-atm layer
> > IF-MIB::ifDescr.72 = STRING: ATM5/0/0.0-atm subif
> > IF-MIB::ifDescr.73 = STRING: ATM5/0/0-aal5 layer
> > IF-MIB::ifDescr.74 = STRING: ATM5/0/0.0-aal5 layer
>
> On the same device as above:
>
> IF-MIB::ifDescr.1 = STRING: GigabitEthernet3/1
> [...]
> IF-MIB::ifDescr.128 = STRING: ATM10/1/0.0-atm subif
> IF-MIB::ifDescr.129 = STRING: ATM10/1/0-aal5 layer
> IF-MIB::ifDescr.130 = STRING: ATM10/1/0.0-aal5 layer
> IF-MIB::ifDescr.131 = STRING: unrouted VLAN 1
>
> > (This may be helpful ... somehow ignore all type 37 and type 49, and sub
> > interface 0 ??)
> > IF-MIB::ifType.70 = INTEGER: sonet(39)
> > IF-MIB::ifType.71 = INTEGER: atm(37)
> > IF-MIB::ifType.72 = INTEGER: atmSubInterface(134)
> > IF-MIB::ifType.73 = INTEGER: aal5(49)
> > IF-MIB::ifType.74 = INTEGER: aal5(49)
>
> But, in my case, I want to graph:
> IF-MIB::ifName.103 = STRING: ATM10/1/0.1-aal5 layer
>
> (it is the only interface which shows the correct traffic)
>
> > I will open a bug if you like.  Is the issue that originated this thread
> > the same as the one I am experiencing?
>
> Well, there are two aspects to the bug:
>
> 1)Devmon should strip invalid characters out of interface names for the RRD
> data sent with the status message.
>
> 2)The Hobbit RRD collector module for devmon should not segfault if data is
> not in the format:
>
> <name_without_spaces> value[:value[:value]]

I fixed this one last night. You can either grab the new do_devmon.c, and run 
'make' again, and copy the hobbitd_rrd over the previous binary, or you can 
grab the new complete patch for a build from scratch:

http://devmon.svn.sourceforge.net/viewvc/devmon?view=rev&revision=90

(at present, viewvc seems to be a bit bust on sourceforge, but these should be 
the URLs to the current versions of the files when it is working:
http://devmon.svn.sourceforge.net/viewvc/devmon/trunk/extras/do_devmon.c
http://devmon.svn.sourceforge.net/viewvc/devmon/trunk/extras/hobbit-4.2.0-
devmon-complete.patch
)

I will update my Hobbit packages with this once I've upgraded my production 
Hobbit box (but the SRPM is in Mandriva cooker already).

Regards,
Buchan