Re: 3.3 Stable Performance Monitoring
We also have a few Solaris machines around. We've purchased a SNMP agent
from Empire Technology (www.empiretech.com) which can report various system
performance related parameters, like swap usage, system load, cpu
utilization, number of open file descriptor, number of processes, etc.
The bad news is that their product doesn't support FreeBSD, although it does
support Linux. So we cannot use this tool to monitor the system
performance. Instead, we need something else which can do roughly the same
thing.
Among so many parameters our immediate interests is the following:
* CPU utilization, % used in Kernel space vs % used in user space
* RAM utilization
* SWAP utilization
* Network bandwidth usage
* number of file descriptors used
As ususal, any hints/comments are more than welcomed. Please do mail a copy
of your response to me directly. Thanks!
I've been writing a program to monitor various values dealing with SNMP --
it's not finished, but it works. Basically, you tell it what to watch,
and if the values go outside defined thresholds or certain values are/are
not met, it triggers an "alert" -- mail, paging (both TAP and SNPP), etc.
Right now, it's running under Linux with ucd-snmp, but porting it over to
FreeBSD should be simple -- the errors I'm getting are dumb ones that are
easily fixed. If anybody's interested, let me know -- it's not available
to the general public (I'm sorta embarresed by the code), but the geeks of
the world can get their hands on what I have so far by asking.
mike
(I'll include one of the config files for your browsing and commentary.)
# Machines at SML
#doublewalk {
#name "r2d2_processlist"
#host "r2d2.smlab.com"
#community "Mlx-20L"
#fromoid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrorflag"
#tooid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrmessage"
#frequency 41
#mode 0
#alert "mike_pager"
#}
#doublewalk {
# name "r2d2_df"
# host "r2d2.smlab.com"
# community "Mlx-20L"
# fromoid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrorflag"
# tooid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrormsg"
# frequency 42
# mode 0
# alert "mike_pager"
#}
doublewalk {
name "palpatine_pslist"
host "palpatine.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrorflag"
tooid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrmessage"
frequency 41
mode 0
alert "mike_pager"
}
doublewalk {
name "palpatine_df"
host "palpatine.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrorflag"
tooid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrormsg"
frequency 42
mode 0
alert "mike_pager"
}
doublewalk {
name "watto_pslist"
host "watto.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrorflag"
tooid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrmessage"
frequency 41
mode 0
alert "mike_pager"
}
doublewalk {
name "watto_df"
host "watto.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrorflag"
tooid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrormsg"
frequency 42
mode 0
alert "mike_pager"
}
# check owen's transmit.LOCK lockfile for the reporting/paging system to make
# sure it's not too long...
reportchain {
name "owen_txlockfile"
host "owenpub.smlab.com"
community "Mlx-20L"
oidroot ".1.3.6.1.4.enterprises.ucdavis.50.101"
frequency 60
alert "mike_pager"
}
doublewalk {
name "owen_processload"
host "owenpub.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.loadtable.laentry.laerrorflag"
tooid ".1.3.6.1.4.enterprises.ucdavis.loadtable.laentry.laerrmessage"
frequency 51
mode 0
alert "mike_pager"
}
doublewalk {
name "owen_df"
host "owenpub.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrorflag"
tooid ".1.3.6.1.4.enterprises.ucdavis.disktable.dskentry.dskerrormsg"
frequency 52
mode 0
alert "mike_pager"
alert "mike_mail"
}
doublewalk {
name "owen_processlist"
host "owenpub.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrorflag"
tooid ".1.3.6.1.4.enterprises.ucdavis.proctable.prentry.prerrmessage"
frequency 53
mode 0
alert "mike_pager"
}
doublewalk {
name "tarkin_processload"
host "www.smlab.com"
community "Mlx-20L"
fromoid ".1.3.6.1.4.enterprises.ucdavis.loadtable.laentry.laerrorflag"
tooid
Re: 3.3 Stable Performance Monitoring
*This message was transferred with a trial version of CommuniGate(tm) Pro* On Sat, 23 Oct 1999, Tom wrote: Another way to the monitor system performance in conjunction to snmp is try to use the MRTG package, in the contrib directory u can find a lot of useful and working plugins that will monitor system performance. *This message was transferred with a trial version of CommuniGate(tm) Pro* On Sat, 23 Oct 1999, Gong Wei wrote: We also have a few Solaris machines around. We've purchased a SNMP agent from Empire Technology (www.empiretech.com) which can report various system performance related parameters, like swap usage, system load, cpu utilization, number of open file descriptor, number of processes, etc. The bad news is that their product doesn't support FreeBSD, although it does support Linux. So we cannot use this tool to monitor the system performance. Instead, we need something else which can do roughly the same thing. Among so many parameters our immediate interests is the following: * CPU utilization, % used in Kernel space vs % used in user space * RAM utilization * SWAP utilization * Network bandwidth usage * number of file descriptors used As ususal, any hints/comments are more than welcomed. Please do mail a copy of your response to me directly. Thanks! The ucd-snmp package includes a snmp daemon (snmpd). That last time I did a snmpwalk on it, it reported lots of stuff like you want. The funny part, is that this server probably works on Solaris too, and doesn't cost anything! BTW, I usually get the network bandwidth off the switch the server is plugged into though. Tom To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
RE: 3.3 Stable Performance Monitoring
Big Brother? home: http://maclawran.ca/bb-dnld/index.html demo: http://maclawran.ca/bb-dnld/new-demo.html -- Duane H. Hesser [EMAIL PROTECTED] To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
Re: 3.3 Stable Performance Monitoring
I'm assuming that you've eliminated the possibility of running this under Linux emulation mode? It may be that the snmp agent diddles too closely with kernel structures to allow for this, but there are emulatable ways to perform each of the tasks listed.. Chuck On Sat, 23 Oct 1999, Gong Wei wrote: Hi all, We have a few 3.3 Stable production servers lying around the campus to serve the university population. As usual, we need some means to manage/monitor the server performance closely. We also have a few Solaris machines around. We've purchased a SNMP agent from Empire Technology (www.empiretech.com) which can report various system performance related parameters, like swap usage, system load, cpu utilization, number of open file descriptor, number of processes, etc. The bad news is that their product doesn't support FreeBSD, although it does support Linux. So we cannot use this tool to monitor the system performance. Instead, we need something else which can do roughly the same thing. Among so many parameters our immediate interests is the following: * CPU utilization, % used in Kernel space vs % used in user space * RAM utilization * SWAP utilization * Network bandwidth usage * number of file descriptors used As ususal, any hints/comments are more than welcomed. Please do mail a copy of your response to me directly. Thanks! To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
Re: 3.3 Stable Performance Monitoring
On Sat, 23 Oct 1999, Gong Wei wrote: We also have a few Solaris machines around. We've purchased a SNMP agent from Empire Technology (www.empiretech.com) which can report various system performance related parameters, like swap usage, system load, cpu utilization, number of open file descriptor, number of processes, etc. The bad news is that their product doesn't support FreeBSD, although it does support Linux. So we cannot use this tool to monitor the system performance. Instead, we need something else which can do roughly the same thing. Among so many parameters our immediate interests is the following: * CPU utilization, % used in Kernel space vs % used in user space * RAM utilization * SWAP utilization * Network bandwidth usage * number of file descriptors used As ususal, any hints/comments are more than welcomed. Please do mail a copy of your response to me directly. Thanks! The ucd-snmp package includes a snmp daemon (snmpd). That last time I did a snmpwalk on it, it reported lots of stuff like you want. The funny part, is that this server probably works on Solaris too, and doesn't cost anything! BTW, I usually get the network bandwidth off the switch the server is plugged into though. Tom To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
Re: 3.3 Stable Performance Monitoring
It turns out one of the principles at Empire Technologies is a friend of mine. I'll send him a note regarding a FreeBSD version and see what comes out of it. - Dave Rivers - To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-stable" in the body of the message
