Cannot connect to ESX 4.0 host after boot

Symptoms

• When an ESX 4.0 host is rebooted, it is not possible to connect to the ESX host using vCenter Server, the vSphere Client or vSphere Web Access for 30-40 minutes.
• When management agents are restarted per , there is a delay of about 12 minutes before it is possible to connect to the ESX host.
• There is a DNS server defined in /etc/resolv.conf on the ESX host, but the DNS server is unreachable. This issue typically occurs on newly installed servers where the network infrastructure is under construction and where DNS is not available.
• The vSphere Client shows this error when attempting to connect directly to the ESX host:
   
  The server could not interpret the communication from the client. (The remote server returned an error: (503) Server Unavailable.)
    
• /var/log/messages logs indicate multiple instances of the vmware-authd service restarting:
   
  Sep 27 11:14:20 esx02 xinetd[2973]: START: vmware-authd pid=21102 from=127.0.0.1 
  Sep 27 11:14:20 esx02 xinetd[2973]: EXIT: vmware-authd status=255 pid=21102 duration=0(sec) 
  Sep 27 11:14:20 esx02 xinetd[2973]: START: vmware-authd pid=21103 from=127.0.0.1 Sep 27 11:14:20 esx02 xinetd[2973]: EXIT: vmware-authd status=255 pid=21103 duration=0(sec) 
   
• /var/log/vmware/hostd.log indicates that the hostdsvc plugin takes a long time to start:
  
  [2009-09-27 11:02:08.815 F664F6D0 info 'App'] Trying blklistsvc 
  [2009-09-27 11:02:08.820 F664F6D0 info 'App'] Trying cimsvc [2009-09-27 11:02:08.831 F664F6D0 info 'App'] Trying directorysvc 
  [2009-09-27 11:02:08.845 F664F6D0 info 'App'] Trying hostsvc
  [2009-09-27 11:02:20.100 F664F6D0 info 'App'] Trying httpnfcsvc[2009-09-27 11:02:20.100 F664F6D0 verbose 'App'] [DiskLibWrapperLoader_Init] Trying to load disk lib: diskLibWrapper.so, 
  
  Note: Notice the time difference in each of the line entries to determine the length of time it took for hostsvc to start.

Resolution

Ensure that DNS is configured and reachable from the ESX host. For more information, see .

For troubleshooting purposes, you may wish to login as root to the ESX host with an SSH client and execute this command to observe DNS traffic on port 53:
 
tcpdump -nvvvl -i vswif0 port 53 
 

There may be an issue with the DNS server, if you see output similar to:
 
16:32:26.773870 IP (tos 0x0, ttl 64, id 62337, offset 0, flags [DF], proto: UDP (17), length: 66) 10.21.64.129.58720 > 192.199.99.255.domain: [bad udp cksum af30!] 61766+ PTR? 0.0.0.0.in-addr.arpa. (38)