Branches
Releases
|
Version
|
Focus
|
Date
|
|
8.1.7
|
Major feature enhancements |
13-Nov-2008 19:07 |
|
8.1.6
|
Minor feature enhancements |
01-Oct-2008 19:47 |
|
8.1.5
|
Major feature enhancements |
14-Jul-2008 17:48 |
|
8.1.4
|
Minor feature enhancements |
21-Jun-2008 23:49 |
|
8.1.3
|
Code cleanup |
25-May-2008 12:20 |
|
8.1.2
|
Minor feature enhancements |
16-Mar-2008 13:09 |
|
8.1.1
|
Minor feature enhancements |
10-Feb-2008 21:58 |
|
8.0.4
|
Minor feature enhancements |
06-Jan-2008 13:01 |
|
8.0.3
|
Minor bugfixes |
18-Nov-2007 14:08 |
|
8.0.2
|
Minor feature enhancements |
26-Oct-2007 18:03 |
Articles referencing this project
Comments
[»]
report sar7.16 problem and attach patch for it
by yuyongcong - Aug 2nd 2007 00:30:07
Hello, Mr.Sebastien Godard .
When I used sar7.16,I found a problem.
After adding the parameters "-s 23:59:45 -e 00:00:15",it means
sar will read the records at 23:59:45~00:00:15 from sarfile,but
the result is that sar only read the records at 23:59:45-23:59:55 from
sarfile,the records after 00:00:00 didn't read from sarfile.
I have made a patch to solve the problem.
To repeat this problem,you could do as following:
Step1:run the script
----------------------------------------------------------------------
#!/bin/sh
DATE=`date +%m%d%H%M%S`
date -s 23:59:45
echo "Now time has been set "`date`
echo "Sar starts to write in sar.$DATE "
sar 5 10 -o ./sar.$DATE
hwclock --hctosys
-----------------------------------------------------------------------
After running,it will create a sarfile.
Step2:read created sarfile
Sar Version: sar 7.16
[root@localhost sysstat-7.1.6]# sar -V
sysstat version 7.1.6
(C) Sebastien Godard (sysstat <at> wanadoo.fr)
[root@localhost sysstat-7.1.6]# sar -f testsar/sar.0810000205
Linux (localhost.localdomain) 08/10/2007
23:59:45 CPU %user %nice %system %iowait %steal %idle
23:59:50 all 4.19 0.00 6.59 0.00 0.00 89.22
23:59:55 all 8.02 0.20 11.42 0.00 0.00 80.36
00:00:00 all 5.59 0.00 9.58 0.00 0.00 84.83
00:00:05 all 4.60 0.40 6.60 0.00 0.00 88.40
00:00:10 all 4.61 0.00 6.61 0.00 0.00 88.78
00:00:15 all 3.99 0.20 9.58 2.20 0.00 84.03
00:00:20 all 4.80 0.20 8.20 0.60 0.00 86.20
00:00:25 all 4.39 0.20 6.39 0.00 0.00 89.02
00:00:30 all 4.21 0.00 7.62 0.00 0.00 88.18
00:00:35 all 4.39 0.20 7.39 0.20 0.00 87.82
00:00:40 all 4.20 0.20 7.40 0.00 0.00 88.20
00:00:45 all 7.80 0.00 9.20 1.00 0.00 82.00
Average: all 5.06 0.13 8.05 0.33 0.00 86.42
Created sarfile starts with 23:59:45 ends with 00:00:45
Step3:Problem turn out
[root@localhost sysstat-7.1.6]# sar -f testsar/sar.0810000205 -s 23:59:45
-e 00:00:15
Linux (localhost.localdomain) 08/10/2007
The bug can be fixed with the following patch:
Signed-off-by: "Yu Yongcong" <yuyc@cn.fujitsu.com>
--- sysstat-7.1.6/sar.c.orig 2007-08-14 08:09:28.000000000 +0800
+++ sysstat-7.1.6/sar.c 2007-08-14 10:13:47.000000000 +0800
@@ -840,6 +840,13 @@
set_timestamp(!curr, cur_time[!curr], 16);
/* Set current timestamp */
set_timestamp(curr, cur_time[curr], 16);
+
+ if (tm_start.use && file_stats[!curr].ust_time
+ && (file_stats[curr].ust_time >
file_stats[!curr].ust_time)
+ && (file_stats[curr].hour < file_stats[!curr].hour))
{
+ file_stats[curr].hour += 24;
+ rectime.tm_hour += 24;
+ }
/* Check time (2) */
if (use_tm_start && (datecmp(&rectime, &tm_start) <
0))
@@ -1622,6 +1629,9 @@
((interval < 0) && !from_file[0] &&
!to_file[0]))
set_default_file(&rectime, from_file);
+ if (tm_start.use && tm_end.use && (tm_end.tm_hour <
tm_start.tm_hour))
+ tm_end.tm_hour += 24;
+
/*
* Check option dependencies
*/
Here is the output after the fix:
[root@localhost sysstat-7.1.6]#sar -f testsar/sar.0810000205 -s 23:59:45
-e 00:00:15
Linux (localhost.localdomain) 08/10/2007
23:59:45 CPU %user %nice %system %iowait %steal %idle
23:59:50 all 4.19 0.00 6.59 0.00 0.00 89.22
23:59:55 all 8.02 0.20 11.42 0.00 0.00 80.36
00:00:00 all 5.59 0.00 9.58 0.00 0.00 84.83
00:00:05 all 4.60 0.40 6.60 0.00 0.00 88.40
00:00:10 all 4.61 0.00 6.61 0.00 0.00 88.78
00:00:15 all 3.99 0.20 9.58 2.20 0.00 84.03
Average: all 5.16 0.13 8.40 0.37 0.00 85.94
That's all.
Best Regards.
[reply]
[top]
[»]
Re: report sar7.16 problem and attach patch for it
by Sébastien Godard - Sep 2nd 2007 02:40:24
Thanks for this patch. It has been included in sysstat 8.0.0.
[reply]
[top]
[»]
after all that, sar doesn't work
by Jesse B - Dec 16th 2004 16:33:37
pseries-loaner:/usr/src/packages/SOURCES/sysstat-5.1.3 # ./sar -u 2 5
exec: No such file or directory
dunno whats up, sadc wouldn't start either. sar dies in sa_read right
after trying to read from stdin.
[reply]
[top]
[»]
Re: after all that, sar doesn't work
by Sébastien Godard - Dec 17th 2004 13:26:25
> pseries-loaner:/usr/src/packages/SOURCES/sysstat-5.1.3
> # ./sar -u 2 5
> exec: No such file or directory
>
This is because sar tries to launch its data collector (sadc) and doesn't
find it. Enter "make install" after compiling to install each component in
its directory before using sar.
[reply]
[top]
[»]
fix for spec file
by Jesse B - Dec 16th 2004 13:59:36
I have rpm 4.1 and it complains loudly when trying to package files that
aren't included in the install.
There is also a typo/bug this patch fixes.
here is the patch...
--- /usr/src/packages/SPECS/sysstat-5.1.3.spec 2004-11-19
08:44:06.000000000 -0800
+++ /root/sysstat-5.1.3.spec 2004-12-16 13:57:30.206736256 -0800
@@ -33,7 +33,7 @@ make PREFIX=%{_prefix} \
%install
rm -rf $RPM_BUILD_ROOT
-install -d $RPM_BUILD_ROOT{/var/log/sa}
+install -d $RPM_BUILD_ROOT/var/log/sa
make IGNORE_MAN_GROUP=y \
DESTDIR=$RPM_BUILD_ROOT \
@@ -59,7 +59,7 @@ rm -rf $RPM_BUILD_ROOT
%files
%defattr(644,root,root,755)
-%doc CHANGES CREDITS README *.sample TODO FAQ
+%doc %{_prefix}/doc/sysstat*/*
%attr(755,root,root) %{_bindir}/*
%attr(755,root,root) %{_libdir}/sa/*
%attr(644,root,root) %{_mandir}/man*/*
[reply]
[top]
[»]
compile errors on 2.6
by Jesse B - Dec 16th 2004 11:37:41
Hey, just tried to do an rpmbuild --rebuild on 5.1.3
noticed that the build failed because MAX_BLKDEV isn't defined for 2.6
kernels, for some reason it was removed from major.h.
I defined it to 255 on the command line, and the compile finished. sar
doesn't work, but I have yet to actually install it, so that will probably
help.
[reply]
[top]
[»]
Re: compile errors on 2.6
by Sébastien Godard - Dec 17th 2004 13:30:28
> Hey, just tried to do an rpmbuild
> --rebuild on 5.1.3
>
> noticed that the build failed because
> MAX_BLKDEV isn't defined for 2.6
> kernels, for some reason it was removed
> from major.h.
Yes: MAX_BLKDEV was removed from linux/major.h in 2.6 kernels. Sysstat
5.1.4 has been updated to take this into account.
[reply]
[top]
|
