Cannot access vSphere ESXi host on UCS

[tomytek0709]

I have ESXi 4.1 installed on B200 M2 blades in UCS.  I can ping the IPs of the ESXi hosts from my computer and ping my computer from the ESXi hosts.  I can't connect to any of the ESXi hosts from my computer with the vSphere client.  I cannot even get to the ESXi welcome webpage on any of the hosts.  Has anyone else experience this problem?

[padramas]

Thomas,

Are required TCP ports allowed between your computer network and ESXi hosts ?

http://kb.vmware.com/selfservice/microsites/microsite.do?cmd=displayKC&docType=kc&externalId=1012382&sliceId=1&docTypeID=DT_KB_1_1#VI%20/%20vSphere%20Client

Are management services up and running ? You can try restarting them via console or via SSH session by

services.sh restart

Padma

[tomytek0709]

Thank your for the response.  There is no firewall between UCS and my computer.  When I telnet using with port 80 and 902, I get a response.  I tried restarting management services through the console, but it still does not work.  Is there something I might have to check in UCS Manager?  The host are all reachable IP wise.

[chaausti]

One quick sanity check is to power off your ESXi host and then verify that the pings stop immediatly.

Can you grab a packet capture from the ESX host and verify you see the connection from your desktop?

on the CLI run this command:

esxcfg-vmknic -l

record the name and ip address of the management interface. It should be vmk0, and please upload a screenshot of the output somewhere with your IP addresses removed.

Then run tcpdump-uw -i vmk0 (or whatever interface it is)

When you attempt to connect to the ESX host you should see a 3 way handshake like in this example from an ESXi 5.0 host:

~ # tcpdump-uw -vv -i vmk0  "not port 22"

tcpdump-uw: WARNING: SIOCGIFINDEX: Invalid argument

tcpdump-uw: listening on vmk0, link-type EN10MB (Ethernet), capture size 96 bytes

14:05:40.058467 ARP, Ethernet (len 6), IPv4 (len 4), Request who-has X.X.X.X (XX:XX:XX:XX:XX:XX (oui Unknown)) tell X.X.X.X, length 46

14:05:40.059677 ARP, Ethernet (len 6), IPv4 (len 4), Reply X.X.X.X is-at XX:XX:XX:XX:XX:XX (oui Unknown), length 28

14:05:44.570073 IP (tos 0x0, ttl 126, id 28062, offset 0, flags [DF], proto TCP (6), length 52)

    X.X.X.X.52926 > X.X.X.X.https: Flags [S], cksum 0xb9f4 (correct), seq 2314467615, win 8192, options [mss 1366,nop,wscale 2,nop,nop,sackOK], length 0

14:05:44.570532 IP (tos 0x0, ttl 64, id 20287, offset 0, flags [DF], proto TCP (6), length 52)

    X.X.X.X.https > X.X.X.X.52926: Flags [S.], cksum 0x4219 (incorrect -> 0x8a63), seq 11685832, ack 2314467616, win 65535, options [mss 1366,nop,wscale 9,sackOK,eol], length 0

14:05:44.572293 IP (tos 0x0, ttl 126, id 28063, offset 0, flags [DF], proto TCP (6), length 40)

    X.X.X.X.52926 > X.X.X.X.https: Flags [.], cksum 0x89d0 (correct), seq 1, ack 1, win 16392, length 0

Look for somethign like:

Flags [S]

Flags [S.]

Flags [.]

to access CLI on ESXi press alt-F1

Finally, running tail -f /var/log/messages will show you the logs which may be helpful (/var/log/vmkernel.log on ESXi 5) but if you just installed it it shoul be working.

[tomytek0709]

Thank you for providing a detailed methodology to help troubleshoot this.  Below is a screenshot of the tcpdump when I try to access using http.

 

Below is a tcpdump when I try to access ESXi using the vSphere client.

Below is a screenshot of the log when I try to use https to access ESXi. 

Below is a screenshot of the log when I try to connect using vSphere.

I'm not sure what it means when it shows bytes missing.  Does this look familiar to anyone?