Oracle VM server crashes with disk corruption using SATA 2
613322Dec 17 2007 — edited Jan 7 2008Created OEL 5 x86_64 on a HVM and occassionally the VM server crashes with the following messages on the console:
ata1.0.0 tag 0 cmd 0x25 Emask 0x4 stat 0x40 err 0x0 timeout (frozen)
failed to recover some devices retrying in 5 seconds
comreset failed to respond in 30 seconds
port is too slow to respond, delay is known to occur on vacant sata ports
BMDMA stat 0x24
hardreset failed retry in 5 secs
After this I've no other option but to perform a hard reboot of the system. Down the road I installed oracleasmlib-2.0.3-1.el5.x86_64 and the above problem was persistent when the oel5 virtual machine boots up. I just couldn't get the virtual machine to start. I tried many workarounds but didn't work. I finally disabled SATA2 in the bios and enabled SATA1 and the system has been stable for the past 2 days. Originally my bios was set to SATA1+2.
I guess there's a bug with OEL 5 kernel on VM's using SATA 2 driver.