obfuscurity / nagios-scripts Goto Github PK

Calls of Graphite REST API from https://github.com/obfuscurity/nagios-scripts/blob/master/check_graphite#L95 and
https://github.com/obfuscurity/nagios-scripts/blob/master/check_graphite#L99 are redundant and double the load.

Proposed way:

  response = JSON.parse(RestClient.get(URI.encode(url)))
  if response.empty?
    puts "UNKNOWN no data reported by Graphite"
    exit EXIT_UNKNOWN
  else
    response.each do |cache| 

When you query an inexistent metric in check_graphite, it queries in graphite the following URL:

http://graphite.server/render/?target=inexistent_metric&format=json&from=-5mins

Graphite answers with an empty JSON object:

[]

So JSON.parse(RestClient.get(URI.encode(url))).each doesn't parse anything, and the total value stays the same as initialized:

data["total"] = 0

I think in this case it should throw an exception, as this value will always be 0 and it doesn't make sense to monitor it.

I'm not a ruby developer, so I don't know what is the best way to accomplish this, that's why I'm not providing a pull request...

PD: I'm running graphite 0.9.12, but I think that behavious is common to all graphite versions.

please give examples to sending graphite date to nagios, how it work's and give commands for nagios it should very better

can you see my result and let me know where did i mistake.

viswanatha@viswanatha-desktop:/etc/nagios3$ python check_graphite.py -U http://127.0.0.1/render/?width=586&height=308&_salt=1349434980.323&from=-10minutes&target=carbon.agents.viswanatha-desktop-a.cpuUsage -w 2.25 -c 2.5
[1] 28798
[2] 28799
[3] 28800
[4] 28801
python: can't open file 'check_graphite.py': [Errno 2] No such file or directory
-w: command not found
[2] Done height=308
[1] Exit 2 python check_graphite.py -U http://127.0.0.1/render/?width=586
[2] Done height=308
[3]- Done _salt=1349434980.323
[4]+ Done from=-10minutes

viswanatha@viswanatha-desktop:~/nagios-plugins$ python check_graphite.py -U "http://127.0.0.1:8080/render/?width=586&height=308&_salt=1349434980.323&from=-10minutes&target=carbon.agents.viswanatha-desktop-a.cpuUsage" -w 2.25 -c 2.5
Usage: check_graphite.py [options]

check_graphite.py: error: no such option: -w

I'd like to distribute check_graphite to our hosts, but most of our hosts cannot run it due to various ruby incompatility and package issues. Would it be possible for you to produce a portable version that we can run on amzn_linux and centos boxes?

Hi,

Could you add a function in addition to sum and average that allows for giving thresholds for growth?

A trivial solution could be: threshold (given in percent of starting value) for ending value - starting value. I'm sure there are better solutions.

I just ran some tests with your plugins for integration into the Icinga 2 book and came up with the idea. I'd be happy to help with work on this issue when I have a bit of spare time.

Cheers,
Thomas

Greetings!

I've noticed option "-z", "--zero-on-error", "Return 0 on a graphite 500 error". I am curious why are these internal errors managed explicitly. Is it something check_graphite invokes on API backend or is it just a workaround for known API behaviour?

I am observing "500 Internal errors" on my Graphite installation + check_graphite unpleasantly often. What could be the reason?

Thank you, with regards .zp.

Hi,

Could you add a new function where you can give a value and the thresholds give warning and critical time for when the value is reached. Like value: 95, warning: 2d, critical: 1d to check for percentage of harddisk level.

Maybe I can help with implementing (see #25 for when and why)

Cheers,
Thomas