I'm going to try snmptable command, better to have snmptable command in
mbrowse...
On Mon, Apr 11, 2011 at 12:24 PM, Dave Shield <[email protected]>wrote:
> On 11 April 2011 15:48, Zimmer Hu <[email protected]> wrote:
> >> What was the exact "snmpget" command that you tried?
> >> (And for comparison, what's the full output from snmpwalk?)
> >>
> >
> > Here are the outputs:
> >
> > 1) snmpget:
> > 1a) command line: snmpget ... Table.1.Column1.index
>
> Please can you post the *exact* command that you are using.
> Preferably cut-and-paste from the command line.
>
> Not a generalised version of it - the *exact* command please.
>
>
> > 2) snmpwalk:
> >
> > 2a) CLI: snmpwalk ... Table.1.Column1
> >
> > output: OK. Read all vaule for one column; (but read whole table
> failed
> > with error same as that in 1a))
>
> Again, that's really too vague to be useful.
> Please post the exact command(s) and full output.
>
>
>
>
> > I just can't conceive an example let me go through one row's columns
> first,
> > then move to next row.... Is this to say Snmpv1 is not capable to
> handle
> > table efficiently
>
> It's nothing to do with SNMPv1 vs SNMPv2 - this is a consequence of
> the basic behaviour of the GETNEXT command, combined with the
> OID ordering rules of SMI.
> An "snmpwalk" will *always* traverse tables column by column,
> rather than row-by-row. Regardless of version. That's how SNMP works.
>
>
> > (maybe snmpget ... table.column1, table.column2,..., table.columnX can do
> it)?
>
> "snmpget" is no use here - GET requires you to know the index(es) of
> each row in advance. You'd need to use "snmpgetnext" instead - that
> could indeed take all of the column objects as parameters, and walk
> the table in parallel.
>
> Or else use the "snmptable" command, which does exactly this already.
>
>
> > Is this to say Snmpv1 is not capable to handle table efficiently
>
> SNMPv1 can handle this perfectly sensibly - you just need to
> use the right tool. Have a look at the workings of "snmptable"
> rather than "snmpwalk".
>
>
>
> > I'm implementing my project based on
> > net-snmp table, and I need to return one row after the other in order.
>
> No - you don't. You need to return the results appropriate
> for whatever was in the incoming request.
>
> If the incoming requests are asking for each row in turn,
> then that's what the agent should report.
> If the incoming requests are asking for random elements,
> then *that* is what the agent should report.
> As long as the agent is working correctly, then you'll get
> the right results.
>
> But it's up to the client tools to request the appropriate OIDs,
> and keep track of what to ask for next. SNMP is essentially
> a "stateless" protocol (much the same as HTTP), and the
> agent doesn't need to keep track of who has asked it for what.
>
>
> Dave
>
------------------------------------------------------------------------------
Xperia(TM) PLAY
It's a major breakthrough. An authentic gaming
smartphone on the nation's most reliable network.
And it wants your games.
http://p.sf.net/sfu/verizon-sfdev
_______________________________________________
Net-snmp-coders mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
