Diagnosing client data (vmstat)
list Ye-fee Liang
Two questions: 1. what's the format of [vmstat] ouput that the Hobbit server is expecting? The sar command is not available, so I'm substituting with iostat output that looks like: [vmstat] cpu utilizaton stat using iostat: 0 0 0 0 0 0 0 0 0 0 16 1 0 13 0 0 0 100 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100 0 where the 2nd last tine is the actual iostat output (iostat 300 2) and last line is what I think Hobbit is expecting. Tthe last 4 numbers, 0 0 100 0 are usr, sys, idle and iowait, that I parsed from iostat output. 2. on server side, how do we debug/trace when client data have arrived? There's hist, histlogs, hostdata, logs. I would like to see some reponse to events like: data arriving, data interpreted/extracted, etc. Thank you, Be smarter than spam. See how smart SpamGuard is at giving junk email the boot with the All-new Yahoo! Mail
list Henrik Størner
▸
On Thu, Aug 30, 2007 at 11:39:11AM -0400, ye-fee liang wrote:
Two questions: 1. what's the format of [vmstat] ouput that the Hobbit server is expecting?
Depends on the operating system. Hobbit uses the operating system ID (usually "uname -s" converted to lowercase) to identify what sequence the vmstat columns are in, and based on that feeds the data into the corresponding datasets in the vmstat RRD file.
▸
The sar command is not available, so I'm substituting with iostat output that looks like: [vmstat] cpu utilizaton stat using iostat: 0 0 0 0 0 0 0 0 0 0 16 1 0 13 0 0 0 100 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100 0
Won't work, unless you write code in Hobbit to parse it. What operating system is this ? IRIX, I suppose - it's the only one where the client uses "sar" to collect vmstat data.
2. on server side, how do we debug/trace when client data have arrived?
Login as the hobbit user and run bbcmd hobbitd_channel --channel=client cat This will dump the client-data messages that Hobbit receives. Regards, Henrik
list Ye-fee Liang
Wow! Your response is faster than any commercial service I've seen. Anyway, the OS is SunOS; I'm customizing hobbitclient-sunos.sh, so it's sending iostat output instead of vmstat under the vmstat section. Could I format the output so the server receiving it will think it's coming from vmstat? The vmstat on Sun server does not display the io-wait time, so I substituted using sar, except for this one which does not have sar installed. SunOS server 5.10 Generic_118833-17 sun4v sparc SUNW,Sun-Fire-T200 $ vmstat 2 2 kthr memory page disk faults cpu r b w swap free re mf pi po fr de sr m1 m3 m4 m5 in sy cs us sy id 0 0 0 29045048 14792424 8 30 6 0 0 0 0 0 0 0 0 683 294 694 0 0 100 0 0 0 28957904 14589248 0 5 0 0 0 0 0 0 0 0 0 808 514 797 0 0 100 All new Yahoo! Mail Get news delivered. Enjoy RSS feeds right on your Mail page.
list Trent Melcher
Why dont you use the bb-iostat.sh script on deadcat, probably does some of the stuff you need....it does need to be modified for graphing the numbers. Trent
▸
On Thu, 2007-08-30 at 12:48 -0400, ye-fee liang wrote:Wow! Your response is faster than any commercial service I've seen. Anyway, the OS is SunOS; I'm customizing hobbitclient-sunos.sh, so it's sending iostat output instead of vmstat under the vmstat section. Could I format the output so the server receiving it will think it's coming from vmstat? The vmstat on Sun server does not display the io-wait time, so I substituted using sar, except for this one which does not have sar installed. SunOS server 5.10 Generic_118833-17 sun4v sparc SUNW,Sun-Fire-T200 $ vmstat 2 2 kthr memory page disk faults cpu r b w swap free re mf pi po fr de sr m1 m3 m4 m5 in sy cs us sy id 0 0 0 29045048 14792424 8 30 6 0 0 0 0 0 0 0 0 683 294 694 0 0 100 0 0 0 28957904 14589248 0 5 0 0 0 0 0 0 0 0 0 808 514 797 0 0 100 All new Yahoo! Mail Get news delivered. Enjoy RSS feeds right on your Mail page.
list H. Klomp
Since the topic of vmstat is alreading started I have a system in which de vmstat output is slightly different from the linux expected output. Here, the IDLE and WAIT time are switched. Causing Hobbit to think the wait is is very large and the idle time is very low. (see example) [vmstat] procs memory swap io system cpu r b swpd free buff cache si so bi bo in cs us sy wa id 2 0 164232 34532 67736 44528 0 1 1 4 8 6 1 0 0 5 1 0 164232 33000 67604 45072 0 0 1 29 141 126 3 0 0 97 This is on a hardened linux system. Is there a known os which also has this way of reporting the vmstat info Regards, Bert Klomp
▸
-----Original Message-----
From: Henrik Stoerner [mailto:user-ce4a2c883f75@xymon.invalid]
Sent: donderdag 30 augustus 2007 17:54
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] Diagnosing client data (vmstat)
On Thu, Aug 30, 2007 at 11:39:11AM -0400, ye-fee liang wrote:Two questions: 1. what's the format of [vmstat] ouput that the Hobbit server is expecting?
Depends on the operating system. Hobbit uses the operating system ID (usually "uname -s" converted to lowercase) to identify what sequence the vmstat columns are in, and based on that feeds the data into the corresponding datasets in the vmstat RRD file.
The sar command is not available, so I'm substituting with iostat output that looks like: [vmstat] cpu utilizaton stat using iostat: 0 0 0 0 0 0 0 0 0 0 16 1 0 13 0 0 0 100 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 100 0
Won't work, unless you write code in Hobbit to parse it. What operating system is this ? IRIX, I suppose - it's the only one where the client uses "sar" to collect vmstat data.
2. on server side, how do we debug/trace when client data have arrived?
Login as the hobbit user and run bbcmd hobbitd_channel --channel=client cat This will dump the client-data messages that Hobbit receives. Regards, Henrik
list Henrik Størner
▸
On Fri, Aug 31, 2007 at 12:31:31PM +0200, Klomp, H. wrote:
Since the topic of vmstat is alreading started I have a system in which de vmstat output is slightly different from the linux expected output. Here, the IDLE and WAIT time are switched. Causing Hobbit to think the wait is is very large and the idle time is very low.
Start your Hobbit client on this box with the "--os=rhel3" option. For some perverse reason, there is a vmstat version that does this, and it is used in Red Hat Enterprise Linux 3. Regards, Henrik