Perc3/Di high load on writing (debian 2.4.20)

Ulrich Hochholdinger <[email protected]>
Newsgroups gmane.linux.drivers.aacraid.devel
Message-ID <[email protected]>
Hi,
I experience high load on a PE2650 while writing to disk:

System Setup is:
PE 2650 / 2GB RAM / 2 Xeon 2.4GHz / PERC3/Di with 3x73GB as RAID5
(factory configured)
Kernel is debian 2.4.20-5 / tested this also with 
mem=512M to exclude HIGHIO

Also tested on an older PE2600 2xPIII / 1G RAM / Perc3/Di 3 Disks as RAID5
with Debian kernel 2.4.18

With the command
dd if=/dev/zero of=bigfile bs=1024 count=8000000)
I start writing a big file to disk, and system load starts to increase.
(this has been tested on ext2 and ext3 fs with same result)

Watched the stats via "vmstat 1" shows that I receive about additional 800 irq/s
while this process is running, looking in /proc/interrupts these interrupts 
nearly all belongs to the aacraid.

Another thing is that there are about 8 to 10 noninterruptible sleep
processes.
Since the 2.4.20 has the patch for the irq storm in rx.c (the
time_before problem) there must be another problem with the aacraid driver.

----
vmstat 1 (before this massive write:)

procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0  0   1888 244104 314680 1293220    0    0     0    32  139    87  0 0 100  0
 0  0   1888 244104 314680 1293220    0    0     0     0  168   201  0 0 100  0
 0  0   1888 244104 314680 1293220    0    0     0     0  127    66  0 0 100  0
 0  0   1888 244104 314680 1293220    0    0     0     0  151   152  0 0 100  0
 0  0   1888 243796 314684 1293524    0    0     0   992  647   812  0 0 100  0
 0  0   1888 243792 314688 1293524    0    0     0   192  213   216  0 0 100  0
 0  0   1888 243792 314688 1293524    0    0     0     0  127    52  0 0 100  0
 0  0   1888 243756 314692 1293556    0    0     4   248  392   801  0 0 100  0
 0  0   1888 243724 314692 1293588    0    0     0   464  360   612  0 0 100  0

------
vmstat 1 (while this massive write:)

procs -----------memory---------- ---swap-- -----io---- --system-- ----cpu----
 r  b   swpd   free   buff  cache   si   so    bi    bo   in    cs us sy id wa
 0 12   2572  10904 202788 1690004    0    0     0 27208  963   544  0 12 88  0
 1 11   2700  10964 199104 1695008    0    0     0 25796  959   472  0 12 88  0
 1 12   2956  10160 197152 1700420    0    0     0 27176  963   413  0 12 87  0
 0 12   2956  11144 192820 1703060    0  948     0 28404  962   513  0 11 88  0
 1 10   3084   9844 192372 1705572    0    0     0 26848  947   464  0 15 85  0
 1  7   3084  10496 188124 1711328    0    0    12 27740 1106   660  0 13 86  0
 0  9   3084  12372 182596 1716168    0    0     0 26004  938   489  0 12 87  0

Since this machine is not fully productive until next week, I can test different
kernel/patches and so on if someone has an idee.

Gruss
	Ulli

-- 
\ Ulli Hochholdinger                               E-Mail: [email protected] \
/ ++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++ /
\ Sometimes I think the surest sign that intelligent life exists elsewhere \
/ in the universe is that none of it has tried to contact us. (Calvin)     /

_______________________________________________
Linux-aacraid-devel mailing list
[email protected]
http://lists.us.dell.com/mailman/listinfo/linux-aacraid-devel
Please read the FAQ at http://lists.us.dell.com/faq or search the list archives at http://lists.us.dell.com/htdig/
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.