We run systems which have built in support for IPMI to essentially restart themselves when they lock up.
We had an odd situation with a server where discs ran at half the speed when one of the redundant power supplies failed (rated 750w each with max utilisation reported at 372w) and Ceph was kicking the discs out and in constantly and hasn't been tuned yet to restrict recovery.
Some of our virtuals subsequently had their disc I/O starved and just sat there, essentially hung with no network.
Would be great if Proxmox added emulated IPMI for KVM to reset the virtual guest when it's watchdog timer expired. This appears to be part of Qemu already and hopefully requires no custom development.
Reference:
https://www.mankier.com/1/qemu#-device_1
-device ipmi-bmc-sim,id=id[,slave_addr=val]
Add an IPMI BMC. This is a simulation of a hardware management interface processor that normally sits on a system. It provides a watchdog and the ability to reset and power control the system. You need to connect this to an IPMI interface to make it useful
The IPMI slave address to use for the BMC. The default is 0x20. This address is the BMC's address on the I2C network of management controllers. If you don't know what this means, it is safe to ignore it.
We had an odd situation with a server where discs ran at half the speed when one of the redundant power supplies failed (rated 750w each with max utilisation reported at 372w) and Ceph was kicking the discs out and in constantly and hasn't been tuned yet to restrict recovery.
Some of our virtuals subsequently had their disc I/O starved and just sat there, essentially hung with no network.
Would be great if Proxmox added emulated IPMI for KVM to reset the virtual guest when it's watchdog timer expired. This appears to be part of Qemu already and hopefully requires no custom development.
Reference:
https://www.mankier.com/1/qemu#-device_1
-device ipmi-bmc-sim,id=id[,slave_addr=val]
Add an IPMI BMC. This is a simulation of a hardware management interface processor that normally sits on a system. It provides a watchdog and the ability to reset and power control the system. You need to connect this to an IPMI interface to make it useful
The IPMI slave address to use for the BMC. The default is 0x20. This address is the BMC's address on the I2C network of management controllers. If you don't know what this means, it is safe to ignore it.
