|
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] Using kexec to reboot a dom0
Hello, I would like to use kexec to fast-reboot a dom0, but I can't make it working properly. I'm using stock Xen 4.20 from Debian 13, stock kernel 6.12, and kexec-tools 2.0.29. I am also only running PV guests, but I'm not sure why it would have an impact since the kexec procedure runs once every domU has been shut down and all Xen services stopped. The best I have right now is : kexec --load -t multiboot2-x86 \ /boot/xen \ --command-line 'smt=off sched-gran=core dom0_mem=512M,max:512M com2=57600,8n1,0x2f8,3 console=com2,vga' \ --module '/vmlinuz root=UUID=... ro console=hvc0 net.ifnames=0 console=tty0 console=ttyS1,115200n8' \ --module '/initrd.img' systemctl kexec ... which immediatly prints on the serial console : (XEN) Executing kexec image on cpu0 Then nothing happens for 3min30s and finally the new Xen payload starts, and everything boot fine. Obviously that delay is the most problematic thing and I wonder how to debug this (what shoud be more verbose ?). I do capture every Xen and kernel messages through the serial console (through a BMC then IPMI). I also ran in those problems : - the "-t multiboot-x86" would not work, I got the "Executing kexec image on cpu0" message but the server was definitively hung (waited 6min) - giving kexec the /boot/xen-4.20-amd64.gz payload which Grub boots fine would fail with "Cannot determine the file type of /boot/xen-4.20-amd64.gz" ; it only worked once I decompressed the Xen payload beforehand Has anyone successfully used kexec on dom0 and could help ? Thanks ! PS: I use kexec on the same hardware successfully without Xen (full reboot in 30 to 45s), there does not seem to be any BIOS/firmware/driver issue.
|
![]() |
Lists.xenproject.org is hosted with RackSpace, monitoring our |