I'm not really sure what happened here, but kernel logfile and serial
console log said this:
__alloc_pages: 0-order allocation failed.
__alloc_pages: 0-order allocation failed.
__alloc_pages: 0-order allocation failed.
__alloc_pages: 0-order allocation failed.
eth0: memory shortage
__alloc_pages: 0-order allocation failed.
__alloc_pages: 0-order allocation failed.
__alloc_pages: 0-order allocation failed.
...
System was unstable and hard-locked when I did a sysrq-p. syslog died when
the error showed up, so no data is available. There's 1GB in this system
with Apache (128 children) and MySQL (192MB keycache) running.
I'm not sure how to recreate the error. System was running fine for 5 days
before it showed up.
-Vibol