Solved Tools to monitoring system?

  • Thread starter Deleted member 66267
  • Start date
D

Deleted member 66267

Guest
I only know how to use top (basic level) thanks to the knowledge I had with Linux. But I heard people said running top itself will have impact on the performance, it will affect the very performance it's monitoring so it shouldn't be used for serious performance monitoring. So what to use now?
 
top on my system draws .08% on CPU0 according to top.
I really don't know anything lighter.
I don't know. I heard people said that top affects performance.
BTW, top on my system is not as lightweight as yours. Do you run top when running heavy jobs like compiling a large source code in parallel? Sometimes I used too much make jobs. Using top, I could see how much clang++ processes it spawned and the resources it used so I could adjust it to a more reasonable value.
 
BTW, do you know if top monitor the system performance in real time? I see there is delays in the output of top.
 
ANY monitoring tool running on the very same CPU that it monitors will affect it's subject... There is no Schroedinger's cat, neither in base nor in sysutils/*. You may want to have a look at systat(1), vmstat(8), accton(8), net/vnstat, sysutils/bpytop, psearch -c sysutils -s monitor|most
How to interpret the output of vmstat? It's obscure for me. At least top is more user friendly. BTW, do you know how to exit systat? I can only quit it using Ctrl+C.
 
Do not read the manual pages, TL;DR. quit Exit systat. (This may be abbreviated to q.)
Not work, man. The 'q' letter is commonly used across tools like top, less,... I did try to use 'q' to exit systat but it didn't work. Only Ctrl+C works.
 
That's because you didn't RTFM as per my request:
Code:
     Certain characters cause immediate action by systat.  These are

     ^L          Refresh the screen.

     ^G          Print the name of the current ``display'' being shown in the
                 lower window and the refresh interval.

     :           Move the cursor to the command line and interpret the input
                 line typed as a command.  While entering a command the
                 current character erase, word erase, and line kill characters
                 may be used.

     The following commands are interpreted by the ``global'' command
     interpreter.
     [...]
If there's any other manual page I should read to you, just drop me a note (by PM). That's the true spirit of the almighty BeaSD - the power to serve.
 
That's because you didn't RTFM as per my request:
Code:
     Certain characters cause immediate action by systat.  These are

     ^L          Refresh the screen.

     ^G          Print the name of the current ``display'' being shown in the
                 lower window and the refresh interval.

     :           Move the cursor to the command line and interpret the input
                 line typed as a command.  While entering a command the
                 current character erase, word erase, and line kill characters
                 may be used.

     The following commands are interpreted by the ``global'' command
     interpreter.
     [...]
If there's any other manual page I should read to you, just drop me a note (by PM). That's the true spirit of the almighty BeaSD - the power to serve.
So it uses the same syntax as vi, have to type : before actual command. Thanks.

BTW, systat provides too little information.

Code:
                    /0   /1   /2   /3   /4   /5   /6   /7   /8   /9   /10
     Load Average   ||||

                    /0%  /10  /20  /30  /40  /50  /60  /70  /80  /90  /100
root           idle XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
root           idle XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
root           idle XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
root           idle XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX
failure     firefox X
failure     firefox X
root           Xorg X
It's almost useless compared to top.
 
alias hgrep history|bsdgrep
Code:
paul@t450s:~ $ hgrep systat
   134  0:17    systat -vmstat
   213  0:43    systat -ifstat 10
  1351  13:17   systat 3
  2300  16:39   systat -zarc
  2618  6:46    systat -3
  2619  6:46    systat -netstat -3
  2620  6:46    systat -iostat -3
  2622  6:47    systat -pigs -3
  2623  6:47    systat -pigs 3
  2624  6:47    systat -ip4 3
  3213  20:47   systat -ip 3
  3214  20:48   systat -vmstat -3
  3216  20:49   systat -vmstat 4
  3217  20:49   systat -vmstat 3
  4301  4:25    man systat
  4329  7:15    hgrep systat
 
Some people prefer sysutils/htop over top. Monitoring "the system" is a broad term.
I have sysutils/htop for infrequent use. On desktop I prefer sysutils/mate-system-monitor.

For GPU, I have sysutils/radeontop. Works on AMD only.

1616136839143.png
 
For longterm monitoring you can use SNMP.
It require SNMP server + collector.

FreeBSD has builtin SNMP daemon bsnmpd().
For advanced purposes you can install net-mgmt/net-snmp from ports/packages.
Both programs are easy to run.

There is a lot of snmp collectors exsist.
I know about perfect collectors even for windows-os.
For simple tasks you may install net-mgmt/mrtg, it draws charts from SNMP-data or custom shell scripts.
Mrtg graphs are generated as simple webpage, so your performance graphs may be accessible via http web browser.
A dedicated external server for SNMP monitoring gives additional benefits.
 
Sure you will. But you're going to need to use something like Zabbix or Munin, those tools are probably a bit overkill for 'casual' monitoring.
yes, Zabbix is a complete monster, I use it for servers,network,and anything that have SMTP
but I reffer to a console tool in realtime
 
What about console tool gstat(8)?
The gstat utility can be used to monitor I/O transactions of geom(4) devices.

So gstat perfectly shows I/O load of disks, partitions, geom*.
It is very useful tool for performance overview of storage subsystem.
 
Back
Top