Today I encountered a weird phenomenon which I did not encounter yet. I installed the VMware Converter Plugin for VirtualCenter client, so far so good. Only thing left to do was to tick the checkbox to enable the plugin. But as soon as I ticked it the VirtualCenter client crashed without any warning. I rebooted the server, and tried it again… without any luck unfortunately. I reinstalled the plugin, again no luck… I checked the wonderful VMTN forum and someone already solved this problem, it’s due to the “ssleay32.dll” in the system32 directory which is incompatible with the one VMware converter uses. Remove it, and everything will work like it should! If this doesn’t work, check out this follow-up.
Server
New HA error messages
Just noticed the new errors HA produces at the console when a host is isolated in ESX 3.5:
Node isolated from network. Either repair network or override agent startup using option -noiso
And when the host returns from isolation:
Node no longer isolated, will restart
This makes troubleshooting a lot easier! Thanks VMware 🙂
Queue depth and alike settings lost after an upgrade to ESX 3.5
I just had a discussion about an upgrade from 3.0.x to 3.5. During that conversation the new HBA drivers that VMware uses came up. After an upgrade it seems that the queue depth settings are lost. Thinking about it I can understand why. ESX 3.5 uses different driver for most HBA’s. For instance Emulex, their driver for the LP11002 used to be the “lpfcdd_732” and now is the “lpfc_740”. After an upgrade the new driver/module doesn’t get the options you provided with “esxcfg-module” because it was specified for that specific “old” module. This means that you have to set these options again!
CDP, why?
That’s what I thought at first, why do I need CDP(Cisco Discovery Protocol)? Well yesterday CDP proved to be an excellent option for troubleshooting weird networking problems. One of my customers replaced a faulty NIC and accidentally switched two network cables. One of these cables was part of an etherchannel, this caused all kind of strange routing / connection errors to the virtual machines. CDP was very useful because I could setup a remote connection to the customer and check which vmnic was connected to which port on the switch. So use CDP when in doubt!
Click on the blue cloud next to the vmnic name on the right:
This is the result:
(Screenshots taken from the RTFM Upgrade Guide.)
New version of Virtual Iron will be released later this day
A new version of Virtual Iron will be released later this day:
Release Number: 4.2.13
New Feature:
- Linux VSTools installation for kernel updates / new distros
Fixes:
- Windows VSTools hangs during installation
- SLES 10 CD Based installs fail
- Archive manager stops running
- Network associations lost when ports are reset
- When VSTools is checked on W2k3 domain controller guest all logins fail