snmpv3: rwuser/rouser

[redhouse101]

Hi, everybody,

In snmpd.conf manpage, I see rocommunity6/rwcommunity6 is configured for
using IPV6. For SNMPv3, why there is no rouser6/rwuser6? There is only
rouser/rwuser for SNMPv3.

rouser/rwuser is a wrapper from com2sec. Com2sec6 does exist. Should we have
rouser6/rwuser6? I can't find it in code either.


Thanks,

RH

[Dave Shield]

Post by redhouse101
In snmpd.conf manpage, I see rocommunity6/rwcommunity6 is configured for
using IPV6. For SNMPv3, why there is no rouser6/rwuser6?

Because SNMPv3 requests aren't subject to source restrictions.
It doesn't matter where the incoming request was sent from.
All that's important is whether the user/password(s) are correct.

Post by redhouse101
rouser/rwuser is a wrapper from com2sec.

No.
This kicks in at the next level (group) - not com2sec.

Dave

[Mark Atwood]

I'm using 5.3 and I generated a mfd project, along with it's
autogenerated AgentX main and it's Makefile. I had hoped/assumed
that compiling that "as is" would at least compile, even if it
doesn't do anything. It doesn't.

In fooBarTable_data_access.c, in the function fooBarTable_container_load,
there are chunks of code that look like this:

/*
* make sure there is enough space for fooBarThingName data
*/
if ((NULL == rowreq_ctx->data.fooBarThingName) ||
(rowreq_ctx->data.fooBarThingName_len <
(fooBarThingName_len * sizeof(fooBarThingName[0])))) {
snmp_log(LOG_ERR, "not enough space for value\n");
return MFD_ERROR;
}
rowreq_ctx->data.fooBarThingName_len =
fooBarThingName_len * sizeof(fooBarThingName[0]);
memcpy(rowreq_ctx->data.fooBarThingName, fooBarThingName,
fooBarThingName_len * sizeof(fooBarThingName[0]));


It looks like these are being generated by mib2c-conf.d/generic-ctx-get.m2i

They don't compile, giving an error that looks like

rdbmsDbTable_data_access.c:321: error: ‘fooBarThingName_len’ undeclared (first use in this function)
rdbmsDbTable_data_access.c:321: error: ‘fooBarThingName’ undeclared (first use in this function)

which is not surprising, since "fooBarThingName_len" and
"fooBarThingName" are members of the struct , not local or global
variables that can be referred to like this. This can't compile.

I've tried various permutations of mib2c mfd selections, and this
problem remains.

Is this a bug/oversight in the mib2c mfd conf, or what?

Is anyone actually using mib2c mfd for 5.3?