UNSOLVED

ftomek

updated

10 years ago

F

ftomek

3 Posts

0

1914

May 26th, 2016 08:00

N4000 interstack LACP communicatins issue

I have 2 stacks N4032F (interstack links 4x10GE) connected by 4x10GE LACP mode active. This is core. Some Cisco switches are connected to them on LACP 2x10GE

I can not ping some hosts from one side to another. This issue is observed when I test from Dell, but also host to host.

I noticed that if is not pingable ARP adge grows. It looks like not ARP are transfered by LACP. All tested hosts are in the same VLAN. Dell switches have interfaces in that vlan where pinged hosts are. When I added static ARP I can ping. When I remove static after some minuts communication breaks. It is observed only pinging some hosts located on Microsoft HV.

----172.16.0.150 PING statistics----
4 packets transmitted, 0 packets received, 100% packet loss
round-trip (msec) min/avg/max = <10/<10/<10

N4032F_1_DC1#show arp

Static ARP entries are only active
when the IP address is reachable on a local subnet


Age Time (seconds)............................. 1200
Response Time (seconds)........................ 1
Retries........................................ 4
Cache Size..................................... 4096
Dynamic Renew Mode ............................ Enable
Total Entry Count Current / Peak .............. 71 / 105
Static Entry Count Configured / Active / Max .. 0 / 0 / 128

IP Address       MAC Address        Interface       Type      Age
---------------  -----------------  --------------  --------  -----------
172.16.0.2       F8B1.567D.8D9E     Vl1             Local         n/a
172.16.0.3       F8B1.567D.8D9A     Vl1             Dynamic    0h  1m  2s
172.16.0.4       085B.0EBD.7896     Vl1             Dynamic    0h  0m  2s
172.16.0.100     0015.5D05.1106     Vl1             Dynamic    0h  0m  0s
172.16.0.111     A4BA.DBF8.B1BF     Vl1             Dynamic    0h  0m  6s
172.16.0.150     0015.5D05.1103     Vl1             Dynamic    0h  1m  5s
























I made some tests. First I closed on of legs on LACK to Cisco switches.

Then I started to test stack1 to stack2 4x10GE link closing links. When I had only one link communications works fine. When I had 2 links do different stack members the problem persist, when I had 2 links connecting each stack member horisontally in seems to be OK.

Does anybody have such problems? How can I solve this issu. I need 4x10 interstack link. I havent test 2x2 horizontaly, because at the moment I can only do it remotely.

  • ftomek

    3 Posts

    1190

    0

    Posted May 31st, 2016 08:00

    Thanks for your answer.

    I will verify you soution. But in the meantime I switched off buggy stack and returned to separated switches. I was working network and I had no more time for experiments.

  • bealdrid

    1 Rookie

    •

    56 Posts

    1190

    0

    Posted May 31st, 2016 08:00

    This sounds similar to a problem we were having on our N4032Fs.  Hosts directly connected to our N4032F stack could not ping across to some hosts in our M1000e blade center with M8024K switches that are stacked.  It seemed like some sort of stacking/cross switch port channel bug.  This also seemed to happen right after we updated firmware to 6.3.0.3 (which was later pulled from the download site).  We changed the hashing algorithm for the port channels on the N4032 stack to "3", which I believe is Source IP, and this solved it.  By default the algorithm is set to "7" which is "enhanced".  No idea whether this will work for you though.  Hope this helps.