Messages in this thread | | | Date | Sun, 6 Jul 2008 08:42:59 -0400 (EDT) | From | Justin Piszcz <> | Subject | Re: Lots of con-current I/O = resets SATA link? (2.6.25.10) |
| |
On Sun, 6 Jul 2008, Justin Piszcz wrote: > On Sun, 6 Jul 2008, Justin Piszcz wrote: >> On Sat, 5 Jul 2008, Justin Piszcz wrote: >>> On Sat, 5 Jul 2008, Robert Hancock wrote: >> After the rebuild is completed I will perform more testing.
The following setup is in-place:
Kernel: 2.6.25.10 Boot options: none NCQ: disabled on all disks irqbalance: disabled/not running
Removal of all old test data:
$ for i in [0-9]*; do rm -rf $i & done .. [468] 5158 [469] 5159 [470] 5160 [471] 5161 [472] 5162 $
-- no problems yet --
run untar test+copy test:
$ mkdir `seq 0 20` $ ./runtest.sh
for i in `seq 0 20` do cd $i tar xf ../linux-2.6.25.10.tar & cd .. done
for b in `seq 21 40` do cp linux-2.6.25.10.tar $b & done
-- waiting for problems --
If no problems at the end of this test, will start up IRQ balance (how I normally ran my system) and re-test to see it is relegated to the irqbalance daemon running.
-----------------------------
Success ..
$ du -sh `seq 0 40` 311M 0 311M 1 311M 2 311M 3 311M 4 311M 5 311M 6 311M 7 311M 8 311M 9 311M 10 311M 11 311M 12 311M 13 311M 14 311M 15 311M 16 311M 17 311M 18 311M 19 311M 20 272M 21 272M 22 272M 23 272M 24 272M 25 272M 26 272M 27 272M 28 272M 29 272M 30 272M 31 272M 32 272M 33 272M 34 272M 35 272M 36 272M 37 272M 38 272M 39 272M 40
-----------------------------
.. no problems .. .. proceeded with test mentioned above ..
Second round:
The following setup is in-place:
Kernel: 2.6.25.10 Boot options: none NCQ: disabled on all disks irqbalance: enabled
--
Remove old cruft:
$ for i in `seq 0 40` > do > rm -rf $i & > done
Create new directories, re-run test:
$ mkdir `seq 0 20`
$ ./runtest.sh
.. no problems either ..
--
I do not have a method to reproduce the problem at will unfortunately.
Justin.
| |