2 Replies Latest reply on Dec 22, 2018 12:57 AM by ManivelR

    VSAN Network with DVS IP hash issue

    ManivelR Enthusiast

      Hi Guys,


      VSAN Network with DVS IP hash issue



      Im using 2 * 10 GB NICs for VSAN network.

      DVS with IP hash Load balancing(2 NICs are active/active) and rest all are default option only.There is no Beacon probing enabled.

      In DVS side,we have not created any LAG as of now.



      In physical Nexus Cisco switches side,we have configured static etherchannel(LACP)not dynamic etherchannel.



      Fist 10 G NIC is connected to One Cisco Nexus physical switch and another one 10 G NIC is connected to another Cisco Nexus physical switch.



      Testing phase:-



      When we shutdown any one interface from first physical Cisco Nexus switch(connected with ESXi VMNIC4),traffic is going through another Cisco Nexus switch(connected with ESXi VMNIC5) and it sometimes it is not going through(it is intermittent issue) and vice versa.



      When we shudown both the interfaces from 2 Cisco Nexus switchs(VMIC4 and VMNIC5) and enable it back,I keep on getting some Network error as "VSAN cluster paritioned" It is intermittent issue and im able to ping each other ESXi server IP address explicitly through VSAN network.



      To troubleshoot further(as a temporary),when i enable back the DVS LB as "Virtual port id" from IP hash, VSAN cluster paritioned is cleared and everything is showing as green in VSAN cluster Network side(in Monitor tab).



      I dont know where the problem is ? I was reading some ESXi logs to identify but no luck.



      If i introduce LAG for DVS,it will fix this issue ?



      When we use static etherchannel from Physical switch side,LAG is mandatory from DVS side(VMware side) ?



      Any suggestions or ideas ?



      Thank you,

      Manivel R