Hello,
We are planning to moving most of our servers to ESX but before buying our SAN, we want to do some I/O stats to see if iSCSI is enough or if we have to go with FC. So I found a plugin for Nagios that can log I/O stats with iostat. So far it's fine with single disk/one partition servers, but on our Oracle Database 10g server, we have two drives in RAID 1 (/dev/sda) and 4 other drives in RAID 10 (/dev/sdb). When I query iostat, I get :
[root@golgoth ~]# iostat Linux 2.6.9-55.ELsmp (golgoth.acaiq-ctb.lan) 09/21/2009
avg-cpu: %user %nice %sys %iowait %idle 1.83 0.00 3.04 0.79 94.34
Device: tps Blk_read/s Blk_wrtn/s Blk_read Blk_wrtn sda 8.44 12.25 32.40 489731534 1295151278 sda1 0.00 0.00 0.00 21360 7478 sda2 6.60 80.90 44.41 3234128950 1775185432 sda3 0.05 0.20 0.23 7852212 9358056 sda4 0.00 0.00 0.00 2 0 sda5 41.16 38.59 95.20 1542695170 3805566528 sdb 2.84 72.63 28.91 2903208392 1155519494 sdb1 3.67 77.14 23.90 3083762026 955474128 sdb2 1.62 102.92 5.00 4114412086 200045262
So why sda have a lower TPS and bytes read/write than one of it's partitions (sda5, which is the root partition)? I guess I would have to collect the stats for each partitions to get what the total?
Pascal Robert wrote:
Hello,
We are planning to moving most of our servers to ESX but before buying our SAN, we want to do some I/O stats to see if iSCSI is enough or if we have to go with FC. So I found a plugin for Nagios that can log I/O stats with iostat. So far it's fine with single disk/one partition servers, but on our Oracle Database 10g server, we have two drives in RAID 1 (/dev/sda) and 4 other drives in RAID 10 (/dev/sdb). When I query iostat, I get :
I much prefer iostat -x, provides much more information, iostat by itself to me is pretty useless.
I don't even trust i/o stat on my bigger systems I query their performance from the array directly where I can get much better performance information where it really counts.
nate
Hi,
Pascal Robert probert@macti.ca schrieb am 21.09.2009 20:01:20:
[root@golgoth ~]# iostat
I've learned that you must always ignore the first output of iostat but do something along the line of ``iostat 5'' and let it collect some data before even starting to care about it.
Frank.
Am Dienstag, den 22.09.2009, 10:42 +0200 schrieb Frank.Brodbeck@klingel.de:
Hi,
Pascal Robert probert@macti.ca schrieb am 21.09.2009 20:01:20:
[root@golgoth ~]# iostat
I've learned that you must always ignore the first output of iostat but do something along the line of ``iostat 5'' and let it collect some data before even starting to care about it.
Frank. _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Why ignore it, us it correctly. From the manpage:
"The first report generated by the iostat command provides statistics concerning the time since the system was booted"
financial.com AG
Munich head office/Hauptsitz München: Maria-Probst-Str. 19 | 80939 München | Germany Frankfurt branch office/Niederlassung Frankfurt: Messeturm | Friedrich-Ebert-Anlage 49 | 60327 Frankfurt | Germany Management board/Vorstand: Dr. Steffen Boehnert (CEO/Vorsitzender) | Dr. Alexis Eisenhofer | Dr. Yann Samson | Matthias Wiederwach Supervisory board/Aufsichtsrat: Dr. Dr. Ernst zur Linden (chairman/Vorsitzender) Register court/Handelsregister: Munich – HRB 128 972 | Sales tax ID number/St.Nr.: DE205 370 553
Le 09-09-22 à 04:48, Christoph Maser a écrit :
Am Dienstag, den 22.09.2009, 10:42 +0200 schrieb Frank.Brodbeck@klingel.de:
Hi,
Pascal Robert probert@macti.ca schrieb am 21.09.2009 20:01:20:
[root@golgoth ~]# iostat
I've learned that you must always ignore the first output of iostat but do something along the line of ``iostat 5'' and let it collect some data before even starting to care about it.
Frank. _______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Why ignore it, us it correctly. From the manpage:
"The first report generated by the iostat command provides statistics concerning the time since the system was booted"
Ok, but since the Nagios plugin will not call it with an interval, it will be always the same stats. Look like I will have to look at what Dell OpenManage have to offer for I/O stats.
-
Christoph Maser -
Frank.Brodbeck＠klingel.de -
nate -
Pascal Robert