Hi Guys,
I am sorry to bother everybody here. The problem we got is from oracle
9i on AS 2.1 but I found no mail lists specifically for As 2.1 and have
to have a try here.
The problematic system is running an Oracle 9i database on RH AS 2.1
with kernel 2.4.9-e.24enterprise. During peak time, it experiences very
high cpu usage and top shows that several kernel processes are using big
cpu time.
The following lines are from output of top.
top:
27165 oracl 15 0 916M 913M 913M S 0.9 15.1
152:41 oracle
9 root 15 0 0 0 0 SW 0.1 0.0
477:52 bdflush
8 root 15 0 0 0 0 SW 0.0 0.0
604:42 krefilld
6 root 15 0 0 0 0 SW 0.0 0.0
423:01 kswapd
The kernel thread krefilld appears strange to me. It looks like being
related with fs cache flush, swapping and freeing used memory pages,
however the memory usage is not high at all and little swapping activity
either. The following lines are from output of free, sar -B and sar -W
free:
total used free shared buffers cached
Mem: 6180988 6175904 5084 1929228 337732 2739984
-/+ buffers/cache: 3098188 3082800
Swap: 12578672 538236 12040436
sar -W:
10:10:00 AM 0.00 0.00
10:20:00 AM 0.14 0.00
10:30:00 AM 0.01 0.00
10:40:00 AM 0.02 0.00
10:50:00 AM 0.00 0.00
sar -B:
10:10:00 AM 3368.12 1335.97 790526 290046 196156 386294
10:20:00 AM 2697.80 1049.02 791046 291159 192140 386294
10:30:00 AM 1063.50 990.79 801677 252569 220873 386294
10:40:00 AM 4063.40 1330.29 1006899 339327 37619 386294
10:50:00 AM 3475.05 1008.60 956835 281110 147103 386294
I googled on internet and redhat.com but found no helpful information
about this krefilld. A grep on redhat kernel source showed it in
mm/vmscan.c, called by kswapd to move pages from the active list to the
inactive list. The same grep showed nothing in the standard 2.4.26
kernel source. Obviously this is a redhat specific thing.
I also checked redhat errota for kernel updates but found no bugs
related with this krefilld. Since this is a production system and I
won't dare to do any kernel updates without knowing enough information.
Has anybody run into similar situation? Is this a bug already fixed in
later enterprise kernels? Any kernel parameters I can change to solve
the problem?
Your input will be hightly appreciated.
Regards,
Jin
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
This email contains privileged and confidential information intended only for use of the intended recipient and is the property of Qcorps Residential, Inc. Any review, use, distribution or disclosure by others is strictly prohibited. If you are not the intended recipient of this email, please immediately reply to sender and delete all copies of this email along with all attachments. Thank you.
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * * *