domU hung

连接 server-1 这个 domU,出现如下的 error:
# xm console server-1
[31687848.523700] INFO: task jbd2/xvda1-8:175 blocked for more than 120 seconds.
[31687848.523720] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[31687848.523932] INFO: task flush-202:0:666 blocked for more than 120 seconds.
[31687848.523945] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[31687848.524276] INFO: task nagios:15635 blocked for more than 120 seconds.
[31687848.524288] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
[31687848.524424] INFO: task nagios:23634 blocked for more than 120 seconds.
[31687848.524435] "echo 0 > /proc/sys/kernel/hung_task_timeout_secs" disables this message.
...

描述的情况跟这个这篇博客作者描述的基本一致,除了机器型号不同之外。

应该是在比较高的 I/O 延时时才会出现,lucid 的一个 bug

只是单纯的关闭 hung_task_timeout_secs 并没有啥作用,这里提供了一种可行的方式,降低 dirty_ratio 的值。