iostat
list Vernon Everett
Hi all
Searched the list, but never really found anything.
Has anybody managed to get iostat data and graphs working for RedHat linux?
Looking to graph the basics for a system for physical disks.
r/s, w/s, avgrq-sz, await
All the rest is nice to have.
Anybody done this?
Regards
Vernon
NOTICE: This email and any attachments are confidential. They may contain legally privileged information or copyright material. You must not read, copy, use or disclose them without authorisation. If you are not an intended recipient, please contact us at once by return email and then delete both messages and all attachments.
list Stef Coene
▸
On Tuesday 21 October 2008, Everett, Vernon wrote:
Hi all Searched the list, but never really found anything. Has anybody managed to get iostat data and graphs working for RedHat linux? Looking to graph the basics for a system for physical disks. r/s, w/s, avgrq-sz, await All the rest is nice to have. Anybody done this?
I'm doing it for aix. I will adapt my script for linux and let you know the result. For creating and updateing the rrd files, I use the patch from this post: http://www.hswn.dk/hobbiton/2007/01/msg00236.html Stef
list Stef Coene
▸
On Wednesday 05 November 2008, Stef Coene wrote:
On Tuesday 21 October 2008, Everett, Vernon wrote:Hi all Searched the list, but never really found anything. Has anybody managed to get iostat data and graphs working for RedHat linux? Looking to graph the basics for a system for physical disks. r/s, w/s, avgrq-sz, await All the rest is nice to have. Anybody done this?I'm doing it for aix. I will adapt my script for linux and let you know the result.
I think it is working: http://www.docum.org/twiki/bin/view/Hobbit/LinuxIostatGraphs Stef
list Martin Flemming
VERY NICE, Stef !
I've search long time for such a graph !
By the way, is there a possibilty to added other devices into the graph's
.. hope so, think it should be possible, but my script-expertise is not very huge ... :-(
my first shot, change of linux-iostat.pl
< if ( $vmstat =~ /^sd[a-z]/ or $vmstat =~ /^hd[a-z]/ or $vmstat =~ /^md\d/ or $vmstat =~ /^i2o/ ) {
--- if ( $vmstat =~ /^sd[a-z]/ or $vmstat =~ /^hd[a-z]/ or $vmstat =~ /^md\d/ ) {doesn't work .. I've got machines with Adaptec (formerly DPT) ASR-2010S I2O Zero Channel Controller which is recognized as i2o/hda1 i2o/hda3 . . or cciss/c0d0p1 by the way, i search also for such a solaris-script ... Thanks & Cheers, martin
▸
On Wed, 5 Nov 2008, Stef Coene wrote:
On Wednesday 05 November 2008, Stef Coene wrote:On Tuesday 21 October 2008, Everett, Vernon wrote:Hi all Searched the list, but never really found anything. Has anybody managed to get iostat data and graphs working for RedHat linux? Looking to graph the basics for a system for physical disks. r/s, w/s, avgrq-sz, await All the rest is nice to have. Anybody done this?I'm doing it for aix. I will adapt my script for linux and let you know the result.I think it is working: http://www.docum.org/twiki/bin/view/Hobbit/LinuxIostatGraphs Stef
list Martin Flemming
.. little Corecction ... .. the script linux-iostat.pl get now with the change below the data and make the right output with debug-option, but i've got only graphs for e.g hdd or md ... not for i2o/hda1 etc ... 2008-11-06 00:54:09 Using default environment file /usr/lib/hobbit/client/etc/hobbitclient.cfg /usr/lib/hobbit/client/bin/bb 0.0.0.0 "data it-wgs02.trends [iostat-part.i2o/hda5.rrd] DS:tps:GAUGE:600:0:U 054 DS:Kb_read:DERIVE:600:0:U 5312317 DS:Kb_wrtn:DERIVE:600:0:U 1766160 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-part.i2o/hda6.rrd] DS:tps:GAUGE:600:0:U 018 DS:Kb_read:DERIVE:600:0:U 14841 DS:Kb_wrtn:DERIVE:600:0:U 1295592 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-disk.i2o/hda.rrd] DS:tps:GAUGE:600:0:U 13161 DS:Kb_read:DERIVE:600:0:U 1325692846 DS:Kb_wrtn:DERIVE:600:0:U 1162626828 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-part.i2o/hda8.rrd] DS:tps:GAUGE:600:0:U 19835 DS:Kb_read:DERIVE:600:0:U 1273967021 DS:Kb_wrtn:DERIVE:600:0:U 1107176272 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-disk.hdd.rrd] DS:tps:GAUGE:600:0:U 000 DS:Kb_read:DERIVE:600:0:U 0 DS:Kb_wrtn:DERIVE:600:0:U 0 DS:blk_read:DERIVE:600:0:U 0 DS:blk_wrtn:DERIVE:600:0:U 0 [iostat-part.i2o/hda1.rrd] DS:tps:GAUGE:600:0:U 746 DS:Kb_read:DERIVE:600:0:U 38747645 DS:Kb_wrtn:DERIVE:600:0:U 39492064 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-part.i2o/hda2.rrd] DS:tps:GAUGE:600:0:U 000 DS:Kb_read:DERIVE:600:0:U 1136 DS:Kb_wrtn:DERIVE:600:0:U 724 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-part.md0.rrd] DS:tps:GAUGE:600:0:U 000 DS:Kb_read:DERIVE:600:0:U 0 DS:Kb_wrtn:DERIVE:600:0:U 0 DS:blk_read:DERIVE:600:0:U 0 DS:blk_wrtn:DERIVE:600:0:U 0 [iostat-part.i2o/hda7.rrd] DS:tps:GAUGE:600:0:U 200 DS:Kb_read:DERIVE:600:0:U 6478357 DS:Kb_wrtn:DERIVE:600:0:U 11938960 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-part.i2o/hda3.rrd] DS:tps:GAUGE:600:0:U 022 DS:Kb_read:DERIVE:600:0:U 1170025 DS:Kb_wrtn:DERIVE:600:0:U 953480 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U [iostat-part.i2o/hda4.rrd] DS:tps:GAUGE:600:0:U 000 DS:Kb_read:DERIVE:600:0:U 0 DS:Kb_wrtn:DERIVE:600:0:U 0 DS:blk_read:DERIVE:600:0:U DS:blk_wrtn:DERIVE:600:0:U " ... but the graphs are not there (only empty hdd & md0 graphs ... .. something else is missing .... cheers, martin
▸
On Thu, 6 Nov 2008, Martin Flemming wrote:
VERY NICE, Stef ! I've search long time for such a graph ! By the way, is there a possibilty to added other devices into the graph's .. hope so, think it should be possible, but my script-expertise is not very huge ... :-( my first shot, change of linux-iostat.pl < if ( $vmstat =~ /^sd[a-z]/ or $vmstat =~ /^hd[a-z]/ or $vmstat =~ /^md\d/ or $vmstat =~ /^i2o/ ) { ---if ( $vmstat =~ /^sd[a-z]/ or $vmstat =~ /^hd[a-z]/ or $vmstat =~ /^md\d/ ) {doesn't work .. I've got machines with Adaptec (formerly DPT) ASR-2010S I2O Zero Channel Controller which is recognized as i2o/hda1 i2o/hda3 . . or cciss/c0d0p1 by the way, i search also for such a solaris-script ... Thanks & Cheers, martin On Wed, 5 Nov 2008, Stef Coene wrote:On Wednesday 05 November 2008, Stef Coene wrote:On Tuesday 21 October 2008, Everett, Vernon wrote:Hi all Searched the list, but never really found anything. Has anybody managed to get iostat data and graphs working for RedHat linux? Looking to graph the basics for a system for physical disks. r/s, w/s, avgrq-sz, await All the rest is nice to have. Anybody done this?I'm doing it for aix. I will adapt my script for linux and let you know the result.I think it is working: http://www.docum.org/twiki/bin/view/Hobbit/LinuxIostatGraphs Stef
Gruss
Martin Flemming
Martin Flemming
DESY / IT office : Building 2b / 008a
Notkestr. 85 phone : XXX - XXXX - XXXX
22603 Hamburg mail : user-f286aaa49a76@xymon.invalid
list Stef Coene
▸
On Thursday 06 November 2008, Martin Flemming wrote:
.. little Corecction ... .. the script linux-iostat.pl get now with the change below the data and make the right output with debug-option, but i've got only graphs for e.g hdd or md ... not for i2o/hda1 etc ...
That's because there is a "/" in it. You can do s/\//_/g in linux-iostat.pl
to change the / to _.
Or you can strip "i2o/" with s/i2o\///g.
You can do this just before the final command is created (the line where
$ENV{BB} is used):
$rrd_return =~ s/i2o\///g ;
Stef
list Martin Flemming
Great Stuff, Stef !
... three more questions,
a) is it possible to declare some excludes
or something for only active-disks ?
e.g. md0 , the most of my machines have not software-mirror
but some .....
or
/dev/hdd or /dev/cdrom .. in my graph's
it appears as hdd but that's the cdrom and it's empty ....
b)
I've got three empty links in the trend=page
to iostat-disk, how could i delete them or is there a
association with a) ?
c)
I've got negative KB/s on some y-axis ... hmm, why ?
thanks & cheers
▸
martin
On Thu, 6 Nov 2008, Stef Coene wrote:
On Thursday 06 November 2008, Martin Flemming wrote:.. little Corecction ... .. the script linux-iostat.pl get now with the change below the data and make the right output with debug-option, but i've got only graphs for e.g hdd or md ... not for i2o/hda1 etc ...That's because there is a "/" in it. You can do s/\//_/g in linux-iostat.pl to change the / to _. Or you can strip "i2o/" with s/i2o\///g. You can do this just before the final command is created (the line where $ENV{BB} is used): $rrd_return =~ s/i2o\///g ; Stef
list Stef Coene
▸
On Friday 07 November 2008, Martin Flemming wrote:
Great Stuff, Stef !
... three more questions,
a) is it possible to declare some excludes
or something for only active-disks ?
e.g. md0 , the most of my machines have not software-mirror
but some .....
or
/dev/hdd or /dev/cdrom .. in my graph's
it appears as hdd but that's the cdrom and it's empty ....Of course, the script was more a proof of concept. I had the iostat numbers on aix, so it was not that hard to change it for linux. I will update the script and add such feature.
▸
b) I've got three empty links in the trend=page to iostat-disk, how could i delete them or is there a association with a) ?
Euh, can you upload the graph somewhere so I can see it ?
c) I've got negative KB/s on some y-axis ... hmm, why ?
Writes are negative, reads are positive. This is defined in the hobbit-graph. The numbers in the rrd file are fine. I just saw in my script and it also collects the statistics in blocks per second. And by combining the blocks and bytes / second you can get the block size. I can add such graph definitions if anyone is interested.... Maybe I can write a howto based on this script... so anyonce can create a graph from a number. Stef
list Martin Flemming
Hi !
▸
Of course, the script was more a proof of concept. I had the iostat numbers on aix, so it was not that hard to change it for linux.
.. maybe for you .. :-)
I will update the script and add such feature.
Nice to hear .. thanks !
▸
I've got three empty links in the trend=page to iostat-disk, how could i delete them or is there a association with a) ?Euh, can you upload the graph somewhere so I can see it ?
Not really important, maybe my own false configuration, i will answer again, if i've checked it again ...
Writes are negative, reads are positive. This is defined in the hobbit-graph.
Right, it's plausible and documented, thanks ...
▸
I just saw in my script and it also collects the statistics in blocks per second. And by combining the blocks and bytes / second you can get the block size. I can add such graph definitions if anyone is interested....
Yep, my vote, you have ... :-)
▸
Maybe I can write a howto based on this script... so anyonce can create a graph from a number.
I think this will be great !
Cheers & thanks again !
martin
▸
On Fri, 7 Nov 2008, Stef Coene wrote:
On Friday 07 November 2008, Martin Flemming wrote:Great Stuff, Stef ! ... three more questions, a) is it possible to declare some excludes or something for only active-disks ? e.g. md0 , the most of my machines have not software-mirror but some ..... or /dev/hdd or /dev/cdrom .. in my graph's it appears as hdd but that's the cdrom and it's empty ....Of course, the script was more a proof of concept. I had the iostat numbers on aix, so it was not that hard to change it for linux. I will update the script and add such feature.b) I've got three empty links in the trend=page to iostat-disk, how could i delete them or is there a association with a) ?Euh, can you upload the graph somewhere so I can see it ?c) I've got negative KB/s on some y-axis ... hmm, why ?Writes are negative, reads are positive. This is defined in the hobbit-graph. The numbers in the rrd file are fine. I just saw in my script and it also collects the statistics in blocks per second. And by combining the blocks and bytes / second you can get the block size. I can add such graph definitions if anyone is interested.... Maybe I can write a howto based on this script... so anyonce can create a graph from a number. Stef