1

I have a Dell Blade Enclosure with 14 blades and ESXi 5.5 on it. On blade 7 and 14 I have round about 65 % packet loss when I ping the ESXi Management interface. On all other blades there is no packet loss.

The strange thing is, I have this packet loss when I ping from:

blade 1 to blade 7 -> ~ 65% packet loss

but, when I ping from blade 7 to blade 1 at the same time when I ping from blade 1 to blade 7 I have no packet loss at all ... not from 1 to 7 neither from 7 to 1

blade 1 to blade 7 && blade 7 to blade 1 -> 0% packet loss

I have increased the Rx buffer in the ESX CLI but it doesn't help. When I do an esxtop and go to the networktab I don't see any packetdrop:

PORT-ID     USED-BY   TEAM-PNIC   DNAME    PKTTX/s  MbTX/s  PKTRX/s MbRX/s  %DRPTX  %DRPRX
33554433    Management  n/a       vSwitch0  0.00    0.00    0.00    0.00    0.00    0.00
33554434    vmnic0       -        vSwitch0  22.73   0.04    44.50   0.10    0.00    0.00
33554435    Shadow of vmnic0 n/a  vSwitch0  0.00    0.00    0.00    0.00    0.00    0.00
33554436    vmnic1       -        vSwitch0  43.39   0.08    1.91    0.00    0.00    0.00
33554437    Shadow of vmnic1 n/a  vSwitch0  0.00    0.00    0.00    0.00    0.00    0.00
33554438    vmk0       all(2)     vSwitch0  3.66    0.01    1.91    0.00    0.00    0.00

The only thing I see that there are massive interrupts for 0xef:

VECTOR  COUNT/s TIME/int COUNT_0 COUNT_1 COUNT_2 COUNT_3 COUNT_4 COUNT...
0xef    4435.5  1.0      309.2   275.6   402.9   30.5    339.0   15.3 ...

The switch logfile shows me, that the blade network card is flapping sometimes, when I say sometimes I mean once or twice a week for about 1 or 2 minutes.

I don't think that's the reason for that but I haven't any idea anymore what can be the problem. Especially the ping and counter ping situation don't make any sense to me.

Maybe you can help me?

5
  • Which NIC cards are you using? model and brand. Commented Mar 2, 2016 at 14:59
  • I use a broadcom netxtreme II bcm5709s 1000base sx
    – ThoBe
    Commented Mar 2, 2016 at 15:07
  • Without a close analysis, hard to say. I would ask the opinion of the vendor if it is worthwhile to upgrade the firmware of the NIC cards. Commented Mar 2, 2016 at 15:23
  • I'm confused by an inconsistency: "blade 1 to blade 7 -> ~ 65% packet loss ... at the same time when I ping from blade 1 to blade 7 I have no packet loss at all"
    – Jeff Schaller
    Commented Mar 2, 2016 at 15:49
  • When I am only pinging from blade 1 to 7 I have ~65% loss But when I ping from 1 to 7 and! from 7 to 1 I don´t have any loss... neither from 1 to 7 or 7 to 1
    – ThoBe
    Commented Mar 2, 2016 at 15:51

1 Answer 1

1

Well I found the problem and a solution.
I´ve got the packetloss cause of identical MAC-Adresses on different switch ports.
The virtual interface VMK0 Blade 7 had the same MAC dann the hardware interface of Blade 11.

So what I´ve done is, I give the VMK0 interface a new MAC address. For that I go over the iDRAC interface of the Blade Enclousure and loging into the ESXi "GUI". I removed my both networkcards from the Administration interface, restart the Management Network, added them and restart the Management Network again.
So I had some downtime but after that the MAC adress was changed and there was no packetloss anymore.

Thanks to your help.

ThoBe

0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .