Also make sure that all the devices are back lined, in case you have to
raise a RMA. I had a situation where 20 cards in 6500 went faulty after a
DC shutdown activity. Had great difficulty getting all the cards
On Thu, Oct 31, 2013 at 7:14 PM, Andrew Lissitz <all.from.nj_at_gmail.com>wrote:
> I was being a bit sarcastic about the spanning tree question
>
> It still always surprises me now a days when people still rely on it in
> the DC no less. Maybe at the edge you can settle for it because companies
> want to save money, but in the DC? No way. Oh well understood.
>
> Yeah, agreed your products and vendors have a lot to do with the choices
> of what you can and cannot do / the designs you can implement. I would
> however suggest putting this on the 2014 projects list and try and get
> funding for redesign/new gear. Even tweaking STP is not good enough for
> the DC ... STP does not belong in any DC anywhere except for local switch
> loops. Between Cisco, Juniper, Brocade, Avaya, etc Arista coming soon,
> etc a lot of vendors have overlays and or ways to eliminate STP.
>
> A good discussion, please forgive the sarcasm! :)
>
> Andrew
>
> PS - any spooky designs for halloween? Some scary networks out there!
> Lol
>
>
> .
>
> On Oct 30, 2013, at 10:22 PM, James Poplawski <jb.poplawski_at_gmail.com>
> wrote:
>
> > Who uses spanning tree? Lots of people! I have two 3548s as a
> > redundant core and a 3850 with a 10G uplink to each. How do I avoid
> > spanning tree in this instance? How about an antiquated site with two
> > 6500s and no VSS?
> >
> > Sent from my iPhone
> >
> >> On Oct 30, 2013, at 5:36 PM, Andrew Lissitz <all.from.nj_at_gmail.com>
> wrote:
> >>
> >> Hello Old friends! I hope you all are doing well!
> >>
> >> Now a days, who needs spanning tree? You need to run spanning tree
> locally, in case someone plugs the switch into itself, but really do you
> need spanning tree anywhere else?
> >>
> >> Between stacking, LAGs, Virtual Chassis etc . do you really need
> spanning tree anywhere? I might suggest a redesign when you all get a
> chance.
> >>
> >> If L2 is no good for you, (I am anti-L2) then I might suggest you
> consider an overlay technology using L3 MPLS VPNs over GRE or another
> overlay technology; there are quite a few overlay technologies now. I am
> biased towards OpenContrail since it is free, I think it is a good one IMO
> (http://opencontrail.org) and it is GA and most the other solutions are
> still beta. Click the slide show on that link to learn more.
> >>
> >> In your case and with your question, if you direct all traffic to the
> DC that is up, I do not imagine any reason to power cycle your gear in the
> DC which is idle. Why should you? You should be fine. Spanning tree
> will eventually create a loop free environment, but again who uses
> spanning tree today?
> >>
> >> I hope this is a test you are doing, and your friends on this mailing
> list are eagerly awaiting your test results! Do share please!
> >>
> >> Hope all is well!
> >> Andrew CCIE 31840
> >>
> >>
> >>
> >> .
> >>
> >>> On Oct 30, 2013, at 7:48 PM, James Poplawski <jb.poplawski_at_gmail.com>
> wrote:
> >>>
> >>> My thoughts.
> >>>
> >>> Spanning tree will converge to what your running now. If it's running
> >>> in prod and you saved your configs you'll be fine.
> >>>
> >>> The only one I can think of is VSS. Are you running two boxes in east
> >>> coast and two in west coast. Or are you running one in one coast and
> >>> one in the other? VSS can come up in a funky state, which is the
> >>> master versus slave, but it's not going to blow things up. It's just a
> >>> matter of who has the brains (ie running config).
> >>>
> >>> Typically when we have a shutdown, we save configs manually (dump them
> >>> to a laptop/desktop). The only one were mindful of is our WAAS
> >>> appliance because it's running a VB. Everything else is solid. This is
> >>> even more of a precaution of our nightly kiwi backups.
> >>>
> >>> Another worry is working with electrical parts. The time stuff tends
> >>> to break is when the power flips on. Do lightbulb tend to break when
> >>> they're running for 8 hours or when you turn the power on? :)
> >>>
> >>> Another consideration would be BGP. Neighbors that are up the longest
> >>> tends to win via the oldest route resident in best path selection.
> >>>
> >>> Hope this helps buddy, good luck!
> >>> JB
> >>>
> >>> Sent from my iPhone
> >>>
> >>>> On Oct 30, 2013, at 4:07 PM, Cisco Fanatic <ebay_products_at_hotmail.com>
> wrote:
> >>>>
> >>>> ALL -
> >>>>
> >>>> We have 2 data centers across coast to coast. We are power shutting
> down 1
> >>>> data center for 12hrs.
> >>>>
> >>>> We have 6500 (Core) running VSS, Nexus switches, 4500 switches, 3750
> switches
> >>>> to name a few.
> >>>>
> >>>> Now the question is - should we physically shut down the switches and
> >>>> routers?
> >>>>
> >>>> My concern is that once the power comes up and if the switch have to
> be
> >>>> physically rebooted again, the problem is that spanning tree will not
> >>>> reconverge or we might have a network meltdown.
> >>>>
> >>>> What will someone recomemend in this situation?
> >>>>
> >>>> Yuri
> >>>>
> >>>>
> >>>> Blogs and organic groups at http://www.ccie.net
> >>>>
> >>>>
> _______________________________________________________________________
> >>>> Subscription information may be found at:
> >>>> http://www.groupstudy.com/list/CCIELab.html
> >>>
> >>>
> >>> Blogs and organic groups at http://www.ccie.net
> >>>
> >>> _______________________________________________________________________
> >>> Subscription information may be found at:
> >>> http://www.groupstudy.com/list/CCIELab.html
>
>
> Blogs and organic groups at http://www.ccie.net
>
> _______________________________________________________________________
> Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
>
-- Thanks & regards, Nitin Blogs and organic groups at http://www.ccie.netReceived on Fri Nov 01 2013 - 10:53:04 ART
This archive was generated by hypermail 2.2.0 : Wed Jan 01 2014 - 20:26:19 ART