Linux guest cannot randomly connect to another linux guest?

[kenw232]

I'm running ESX 4.0.  I have two Linux guests. One 64bit, and one 32 bit.  The 64 bit guest is a web server which talks to the 32 bit SQL server.  Both work perfectly fine and there has never been a problem.

I recently duplicated the 64 bit guest, gave it a new hostname and IP address and have it connecting to the SQL server too.  But I get connection timeouts once in a while with it.  I can't debug why.  Its works fine for hours, but then two or three timeouts will happen where it simply can't connect to the SQL server.  Yet the SQL server is perfectly fine, even the first 64 bit guest is still using it without a problem.  This connection glitch only happens from the second/new 64bit linux guest. 

Anyone off the top of their head have a clue why this might happen?  The SQL guest is running MySQL, the 64 bit web server is running Apache 2.2.  Eveything else is perfectly fine.  I'm lost.

[GaneshNetworks]

Tell us, how did you duplicate the 64 bit guest?

Make sure that the old and new guests do not have same MAC addresses.

[kenw232]

I replicated it by using VMware converter and converted it to a VMware Workstation VM which I then used to make a bunch of changed like decrease the drive size and whatnot.  Then I Vmware converted it back into ESX server.  Both MAC addresses are different.  Linux ifconfig says "ether 00:0c:29:a8:fd:84" for the one and "ether 00:0c:29:cb:d2:17" for the other.

Any other suggestions would be welcomed.

[lifepo4]

   I have experienced strange problems due to VM cloning, so I always shut down the VM first, then go to *.vmx file and check UUID.location and UUID.bios, Can you make sure both parameters are different for these two Linux servers. If they are the same, I suggest you edit them and save the vmx file, then restart the Linux and see if it solves your problem. FYI, if you make similar change to Windows virtual machine, you would need another CD-Key because Windows now consider this clone to be another computer.

[kenw232]

I SSH'ed into the ESX console and checked both my .vmx files for the two virtual machines.  They are different.

uuid.location = "56 4d 20 b4 36 b3 8d d1-d1 e5 e0 e1 87 a8 fd 84"

uuid.bios = "56 4d 20 b4 36 b3 8d d1-d1 e5 e0 e1 87 a8 fd 84"

uuid.location = "56 4d b9 84 ef 60 a2 ba-07 75 6d 64 dd cb d2 17"

uuid.bios = "56 4d b9 84 ef 60 a2 ba-07 75 6d 64 dd cb d2 17"

Not sure what to look for now...

[lifepo4]

    Unfortunately very often VMware problem is too complicated to figure why it is happening. Did you have this issue the very first time you connect to SQL use this cloned Linux?

    If yes, maybe you can try this

1).  Disconnect the first Linux from SQL, and use only the second Linux, see if you have any issue? If no issue, then somehow the SQL does not like cloned VMs to connect to it at the same time. I had such issue twice (Windows AD account can't be setup to manage vcenter; can't setup VDP,  I have two win2008 server, cloned from each other, one was vcenter, one was DC), either bogged me down for 3-5 days.

2). Is it feasible to set up a new Linux from scratch and try this?  I know this is painful, but.................

    I myself this afternoon got into a situation when cluster was migrated from vDS1 to vDS2, but for one host, vDS1 was persistent and can't be deleted, even after I changed the timeout-for-locked-dvport , it kept telling me WinXP1 was in the defunct dPort, finally I used administrator@vsphere.local to login into vSphere client, and deleted vDS1 from host no.1   :smileyangry:

Good luck

[kenw232]

Thanks, maybe I'll try to install a new VM from scratch.