Load average inflated by +2

[behlendorf]

Thanks, for reporting the issue. This is caused by the txg_quiesce and
txg_sync threads sleeping uninterruptibly. This is just an artifact of
how Linux computes the load average. Those threads really are idle and
your load average is simply overinflated by 2.

I'll look in to getting those two threads to sleep interruptibly so they
don't count toward the load average in the future.

Thanks,
Brian

On Sun, 2011-03-27 at 18:08 -0700, Michael Martin wrote:

Thanks for the ZFS on Linux effort.

I've noticed whenever a pool is loaded, the system load average is
higher than normal and stays there on a pool with no read/write
activity. Basic steps to reproduce:

1. Go to init level 1 to be sure nothing else is causing the load
2. Verify no pools are loaded and note the system load average is
   about 0.05
3. Import the single test pool from a usb 1G flash drive ( zpool
   import test )
4. Watch the load average climb to 2.00 and stay there. cpu's are
   idle and vmstat/iotop confirm very little io.
5. 'zpool export test' and the load slowly drops back down

This is using latest spl git and zfs git on kernel 2.6.38.1 amd64. I
know performance tweaks on on the way, but wanted to mention this
issue. Let me know if I can help debug.

Thanks again for all the work put into this.

[behlendorf]

Fix inflated load average

Kernel threads which sleep uninterruptibly on Linux are marked in the (D)
state. These threads are usually in the process of performing IO and are
thus counted against the load average. The txg_quiesce and txg_sync threads
were always sleeping uninterruptibly and thus inflating the load average.

This change makes them sleep interruptibly. Some care is required however
because these threads may now be woken early by signals. In this case the
callers are all careful to check that the required conditions are met after
waking up. If we're woken early due to a signal they will simply go back
to sleep. In this case these changes are safe.

Closed by bfd214a