tag:blogger.com,1999:blog-3266263034124005485.post8284180942462401563..comments2024-03-24T23:19:30.504+00:00Comments on Fragmentation Needed: Nexus vPC Orphan Portschris margethttp://www.blogger.com/profile/09716555871346949419noreply@blogger.comBlogger16125tag:blogger.com,1999:blog-3266263034124005485.post-36725612030668311742024-03-24T23:19:30.504+00:002024-03-24T23:19:30.504+00:00Server D is configured in active/standby mode, wit...Server D is configured in active/standby mode, with the Active interface cabled to S2.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-72914274960480864012024-03-24T23:14:09.318+00:002024-03-24T23:14:09.318+00:00Why cant d just go to a via s1? And vice versa.Why cant d just go to a via s1? And vice versa.Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-85540187259256220062018-06-15T12:24:35.172+01:002018-06-15T12:24:35.172+01:00Very nice presentation.
I was wandering, what will...Very nice presentation.<br />I was wandering, what will happen for traffic B->C ,C->B if both loose one of their uplinks to the peers and specifically: If B looses uplink to S2 AND C looses uplink to S1.<br />Will then traffic cross the peer link??Anonymoushttps://www.blogger.com/profile/01748583208396317401noreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-88949016954750126922018-02-21T18:07:53.816+00:002018-02-21T18:07:53.816+00:00This is probably the clearest summary on the topic...This is probably the clearest summary on the topic that I have seen. Thanks for the post. Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-37752779967346026102014-10-25T10:11:47.961+01:002014-10-25T10:11:47.961+01:00DL380 Emulex card don't support iscsi with lac...DL380 Emulex card don't support iscsi with lacp, is there another workaround?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-74559964650927288382013-04-15T16:57:09.684+01:002013-04-15T16:57:09.684+01:00Interesting. I haven't noticed this behavior, ...Interesting. I haven't noticed this behavior, but tend to stick to the "don't create orphans" philosophy, so it wouldn't have come up.<br /><br />Is this a bug, or does the behavior make sense for some reason?chris margethttps://www.blogger.com/profile/06646973209424821070noreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-59221706385277352962013-04-15T16:33:43.756+01:002013-04-15T16:33:43.756+01:00Today I found that on Nexus 7000 with NX-OS 5.2(4)...Today I found that on Nexus 7000 with NX-OS 5.2(4) if you create orphan port on secondary vpc peer, the switch will not learn mac address on that orphan port. shahhttps://www.blogger.com/profile/15818883829738651247noreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-73703622530279180492012-07-17T13:54:17.469+01:002012-07-17T13:54:17.469+01:00As an update, VMware vSphere 5.1 will support LACP...As an update, VMware vSphere 5.1 will support LACP on its Virtual Distributed Switch (vDS)! Hooray!Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-76362288995442274522012-06-28T17:17:26.021+01:002012-06-28T17:17:26.021+01:00The 5Ks are dual homed btw. ThanksThe 5Ks are dual homed btw. ThanksAnonymousnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-26927896906455435722012-06-28T17:13:07.297+01:002012-06-28T17:13:07.297+01:00Just curious.
What if it is 2 5Ks connected to 2...Just curious. <br /><br />What if it is 2 5Ks connected to 2 7Ks that are running VPC? The fex are single homed to a 5K each?<br /><br />Will we still face this problem?Anonymousnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-60095918183052565432012-01-30T14:02:01.707+00:002012-01-30T14:02:01.707+00:00Thanks for the post. Perhaps we can add on the lis...Thanks for the post. Perhaps we can add on the list of non-LACP enabled hosts also XenServer. It is possible to enable LACP, but it is not officially supported setup by Citrix, thanks for being corrected if this has changed.<br /><br />As a side note, there is a bug in Linux kernel bonding driver (vanilla kernels up to 3.2) with respect to LACP protocol handling, when even suspended and uni-directional links (switch can't hear LACPDUs from server) are enabled for forwarding. Details at http://marc.info/?l=linux-netdev&m=131852651422444&w=2.netmartinhttps://www.blogger.com/profile/00044790124473770998noreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-83815357182252188582012-01-24T12:54:56.513+00:002012-01-24T12:54:56.513+00:00Thanks for the posts, hope to see more on DC stuff...Thanks for the posts, hope to see more on DC stuff ;)Ariel Liguorihttps://www.blogger.com/profile/11059180104069090363noreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-52958845847029845212012-01-24T11:36:42.819+00:002012-01-24T11:36:42.819+00:00@Sebastien,
Yeah, Solaris with IPMP (pings the ro...@Sebastien,<br /><br />Yeah, Solaris with IPMP (pings the router) and ESX "beacon probe" (requires >2 NICs to be effective) are the schemes I had in mind when I said "most" teaming schemes rely on link state.<br /><br />Any idea how LiveLink works? A cursory googling hasn't turned up clues about the criteria it uses to detect network health. Whether or not it would be able to make a good decision in this case remains an open question in my mind.<br /><br />Either way, these are (probably) slow and (definitely) clunky mechanisms compared to LACP, which takes me back to my last point: Why *wouldn't* you use LACP?chris margethttps://www.blogger.com/profile/09716555871346949419noreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-42469098243544627532012-01-24T00:42:55.359+00:002012-01-24T00:42:55.359+00:00@Jon Still, You're welcome, thanks for taking...@Jon Still, You're welcome, thanks for taking the time to let me know the article was helpful.<br /><br />FEX doesn't change things much<br />- FEX w/ single upstream 5K - we can just think of the FEX ports as 5K ports, nothing changes.<br />- FEX w/ dual upstream 5K - think of the FEX (and attached servers) as working just like "B" and "C" in the example.<br />- FEX w/ dual-Layer vPC - I haven't seen the new software yet, but we're supposed to be able to do vPC FEX with vPC host connections soon. It sounds like the best of both worlds!chris margethttps://www.blogger.com/profile/09716555871346949419noreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-33940001067574720702012-01-23T22:46:05.629+00:002012-01-23T22:46:05.629+00:00Excellent Article. However, for the active/passive...Excellent Article. However, for the active/passive server (D), several constructors develop a tool to ping an address to determine if the network is always reachable via the NIC, though the link status is up. Very usefull for blade server.<br />An example : For broadcom, it is the LiveLinkSebastienhttp://www.cisco.comnoreply@blogger.comtag:blogger.com,1999:blog-3266263034124005485.post-10469119753748539182012-01-23T22:27:59.724+00:002012-01-23T22:27:59.724+00:00Hi Chris,
Great article - really clarified my und...Hi Chris,<br /><br />Great article - really clarified my understanding of vPC peer-link and loop-prevention. I don't see this changing much with server vPC connected through FEXes, but how does FEX vPC change this behaviour?<br /><br />Cheers, and thanks again!<br /><br />Jon. (@xanthein)Jon Stillhttp://www.jonstill.com/noreply@blogger.com