Dear Sun Managers,
I am posting a summary now, even though I only got 3 replies, because
Oracle is aware of it, says it is a bug and says a patch will be out
within a month. Apparently the minute our DBA told the Oracle support
person that the error was occurring every 3 seconds, that reminded him
of this known problem. While one of the respondants is correct that
the problem for now is minor, since it is an anomaly, we wanted to
make sure it was new/standard behaviour for while the problem is minor
now, once this system goes into production we don't want it to baloon
into a serious problem.
I got replies from Kevin Sheehan, Thomas Carter, and one person who's
email I somehow delete (sorry).
I am including the two responses I did get because they can be helpful
for others and are fairly short.
Kevin Sheehan {Consulting Poster Child})
"a) 32 is not a big number
b) pageouts are symptomatic of writes
so my *guess* based on what you say is that Oracle is writing stuff
out every 3 seconds. Probably updating stats or something - the moral
equivalent of a sync(). At 32, you are not dealing with a great deal
of I/O and should probably not be terribly concerned. Thousands of
system calls and a percent or two of system time is almost a Good
Thing when idle."
Thomas Carter:
"This could be due to Oracle regularly writing out cached blocks from
it's memory cache (the SGA). Also, if the SGA isn't big enough, it can
also cause Oracle to constantly free pages to make room for new ones.
Conversely, if it's too big, parts of the SGA get swapped to disk
often. Sizing of the SGA is a difficult parameter to optimally tune."
thanks again to all,
Adam Singer
ORIGINAL POST:
>Dear Sun Managers,
>
>We are seeing some very odd behaviour on a new Oracle server. What is
>happening is that every 3 seconds or so the cpu takes a small hit in
>utilization (no definite idea if it is in user or system process
>space), Page Outs go from 0 to a consistent number (often it was 32)
>and the System Faults (System Calls' trap/interrupt rates (per
>second)) goes way up. Then these values all drop back down for a
>second or two and then again the same results. Oracle 8.0.5 is the
>only app running on this server and this behaviour does not happen
>when oracle is down but I could not find any reference to what is
>going on with this behaviour and frankly I am not sure what is causing
>it. I did not try to truss Oracle but I will try so tomorrow barring
>any other ideas solutions I find.
>
>The server O/S is 2.6, kernel patched to revision 105181-14
>Oracle 8.0.5 (patched by dba), hardware is: Ultra 60, 2 int 9Gb
>mirrored w/ disk suite, RAM is 512MB.
>
>Below is the output of some vmstat's and /usr/ucb/ps -auxwww. A side
>note: the dba run's vmstat at 1 second intervals but doesn't such a
>short time affect what you are measuring? Regardless, the 1 second
>interval is not to fault as we run the same command with oracle down
>and there isn't the "pulse" of page-outs.
>
>Any ideas?
>thanks
>Adam
>
>server% sudo vmstat -S 1
> procs memory page disk faults
>cpu
> r b w swap free si so pi po fr de sr m0 m1 m3 m4 in sy cs
>us sy id
> 0 0 0 27040 36584 0 0 38 5 8 0 0 0 0 0 0 143 5849 33
>1 1 98
> 0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 108 64 37
>0 0 100
>0 0 0 1345464 8080 0 0 0 32 32 0 0 0 0 0 0 156 17694 51
>2 2 96
> 0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 107 45 32
>0 0 100
>0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 112 60 36
>0 0 100
> 0 0 0 1345464 8080 0 0 0 32 32 0 0 0 0 0 0 155 12388 53
>2 1 97
> 0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 107 45 32
>0 0 100
>0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 124 60 36
>0 0 100
> 0 0 0 1345464 8080 0 0 0 32 32 0 0 0 0 0 0 175 16339 59
>2 2 96
> 0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 126 47 34
>0 0 100
>0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 175 60 56
>0 0 100
> 0 0 0 1345464 8080 0 0 0 8 8 0 0 0 0 0 0 144 8327 44
>2 0 98
> 0 0 0 1345464 8080 0 0 0 24 24 0 0 0 0 0 0 122 5730 48
>0 1 98
> 0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 112 62 36
>0 0 100
>0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 108 34 34
>0 0 100
> 0 0 0 1345464 8080 0 0 0 32 32 0 0 0 0 0 0 153 16761 54
>2 2 96
> 0 0 0 1345464 8080 0 0 8 16 16 0 0 0 0 0 0 178 21230 68
>3 2 94
> 0 0 0 1345464 8080 0 0 0 0 0 0 0 0 0 0 0 108 34 38
>0 0 100
>0 0 0 1345464 8080 0 0 0 32 32 0 0 0 0 0 0 152 12655 51
>1 2 97
> procs memory page disk faults
>cpu
> r b w swap free si so pi po fr de sr m0 m1 m3 m4 in sy cs
>us sy id
> 0 0 0 1345464 8080 0 0 0 8 8 0 0 0 1 0 0 130 60 40
>0 0 100
>server% sudo vmstat -S 5 5
> procs memory page disk faults
>cpu
> r b w swap free si so pi po fr de sr m0 m1 m3 m4 in sy cs
>us sy id
> 0 0 0 27112 36584 0 0 38 5 8 0 0 0 0 0 0 143 5849 33
>1 1 98
> 0 0 0 1345456 8072 0 0 0 6 6 0 0 1 0 0 0 162 2750 35
>0 1 98
> 0 0 0 1345456 8072 0 0 0 12 12 0 0 1 0 0 0 190 9334 42
>2 2 97
> 0 0 0 1345456 8072 0 0 0 6 6 0 0 0 0 0 0 116 3135 31
>0 0 99
> 0 0 0 1345456 8072 0 0 0 12 12 0 0 0 0 0 0 132 6178 36
>1 0 98
>
>Sun Microsystems Inc. SunOS 5.6 Generic August 1997
>server% /usr/ucb/ps -auxwww | more
>USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
>oracle 4222 44.0 12.47694462784 ? R 05:55:05 3:58
>ora_lgwr_SCCMS8
>oracle 4564 25.8 13.37819267312 ? S 11:54:53 3:12
>oracleSCCMS8 (LOCAL =NO)
>oracle 4224 6.1 12.47697662832 ? S 05:55:05 9:25
>ora_ckpt_SCCMS8
>oracle 4218 2.1 12.47702462912 ? S 05:55:05 0:20
>ora_dbw0_SCCMS8
>root 4570 0.4 0.3 1808 1520 ? S 12:01:54 0:00
>/usr/local/sbin/sshd
>root 3 0.3 0.0 0 0 ? S Jun 03 27:54 fsflush
>adam 4573 0.1 0.3 1624 1288 pts/0 S 12:01:57 0:00 /bin/ksh
>root 4582 0.1 0.4 2256 2104 pts/0 O 12:02:11 0:00
>/usr/ucb/ps -auxwww
>root 0 0.0 0.0 0 0 ? T Jun 03 0:00 sched
>root 1 0.0 0.1 696 160 ? S Jun 03 0:00 /etc/init
>-
>
>server% /usr/ucb/ps -auxwww | more
>USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
>oracle 4644 1.3 12.47688062752 ? S 12:23:46 0:09
>ora_ckpt_SCCMS8
>root 202 1.1 0.2 1704 944 ? S Jun 03 0:12
>/usr/local/sbin/sshd
>root 3 0.5 0.0 0 0 ? S Jun 03 28:03 fsflush
>root 4687 0.3 0.5 2272 2120 pts/6 O 12:30:03 0:00
>/usr/ucb/ps
>-auxwww
>root 4668 0.1 0.3 1808 1504 ? S 12:28:53 0:00
>/usr/local/sbin/sshd
>
>Sun Microsystems Inc. SunOS 5.6 Generic August 1997
>server% /usr/ucb/ps -auxwww | more
>USER PID %CPU %MEM SZ RSS TT S START TIME COMMAND
>oracle 4644 1.1 12.47688062752 ? S 12:23:46 0:07
>ora_ckpt_SCCMS8
>root 4668 0.6 0.3 1808 1504 ? S 12:28:53 0:00
>/usr/local/sbin/sshd
>root 3 0.3 0.0 0 0 ? S Jun 03 28:02 fsflush
>root 4680 0.3 0.5 2272 2120 pts/6 O 12:29:03 0:00 ps-auxwww
>
>email: regnis@worldnet.att.net
>
email: regnis@worldnet.att.net
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:13:21 CDT