I have a x86 Mikrotik Router 6.22 originally then upgraded to 6.23RC and now 6.23 running on x86 virtual Machine. I have a 4 e1000 nics (MTU 1500) and 1 e1000e nic (MTU 9000) I have configured MPLS and VPLS so I can pass VLANs to my offsite locations running RB750GLs 6 in total. The problem I am having is the x86 host does not crash, log, error anything on screen and I can still login to it but all ethernet adapters go down. This does not happen on any of my other virtual machines and I have tried multiple ideas before posting here. I will quickly summarize what I have tried.
- Re-configure Network Adapters on Mikrotik and ESXi
- Disconnect Ethernet Adapters on ESXi
- Remove all but the single host that has been working for the offsite location
- Disable all other MPLS connections except one to see if a RB750GL is causing the issue.
- Changed MAC Addresses to confirm that their were no conflicts.
- Confirmed MTU is high enough for all locations so its not just silently failing.
- While Host is up and network is working traffic passes fine through all VPLS connections
The only solution after I get my notification from nagios that the host is no longer replying to pings is to just login to the console via ESXi and issue the /system reboot command and when the host reboots the network adapters just start working again.
When MPLS is not set up and running the host stays stable and I have yet to see it crash. The problem is I can’t make it crash on demand which means I get to wait. I have turned on debugging and nothing is visible when that happens.
Does anyone else have any ideas?