[snmp] SNMP v3 discovery bug in 5.1
Josh Bers
jbers at bbn.com
Fri Sep 28 18:45:45 BST 2007
Birgit,
I have confirmed the success of the fix. One strange behavior: On failure
when disconnected, it takes about 10 seconds for the discovery timeout to
happen even though my retry array is [750, 1500, 2250]. Does discovery try
twice per retry?
We would like to get a new release of the stack with this fix in place as
it makes SNMPv3 usable for our application. Can you give me an estimate
for when the next release will be available?
Regards,
Josh
On Tue, 18 Sep 2007, Birgit Arkesteijn wrote:
> Hi Jonathan,
>
> Thanks for trying the new code. I hope that Josh will be able to confirm
> the fix when he returns.
>
> Thanks, Birgit
>
> On 17/09/07 18:03, Jonathan Tung wrote:
> > Hi,
> >
> > Josh is on business travel.
> >
> > I ran his test with the new code, and it seems the bug is fixed, pending
> > verification when he returns.
> >
> > Thanks,
> >
> > Jonathan Tung
> >
> >> Hi Josh,
> >>
> >> I made the two changes to the stack and checked them in to cvs on
> >> SourceForge:
> >> UsmDiscoveryBean.java - new revision: 1.20
> >> SnmpContextv3Basis.java - new revision: 3.14
> >>
> >> I tested that normal SNMPv3 requests still work OK.
> >>
> >> I'm a little uncomfortable running your TestSnmpV3DiscoveryBug.java on
> >> my development system, with bringing interfaces up and down and
> >> (dis)connecting network cables.
> >>
> >> Would you mind testing the changes yourself?
> >>
> >> Thanks, Birgit
> >>
> >> On 17/09/07 14:06, Birgit Arkesteijn wrote:
> >>> Hi Josh,
> >>>
> >>> I agree with you (and Tim) that option 1 seems better, i.e.
> >>> "Move discovery in front of calling super.addPdu that way an exception
> >>> will not result in a tied up transmitter in the context."
> >>>
> >>> As to
> >>> "Why isn't the DiscoveryBean using the context pool mechanism and
> >>> creating its own context directly?"
> >>> I think that's a good point as well. It really should use the pool.
> >>>
> >>> I'll start working on it. However, I'm not sure how much time I'll
> >>> have to change & test the code. Next week I'm off for two weeks.
> >>>
> >>> Oh, and thanks for the good detective work!
> >>>
> >>> Cheers, Birgit
> >>>
> >>
> >
> >
>
>
More information about the snmp
mailing list