brightcoveos / diamond

THIS PROJECT HAS MOVED

Please join us at https://github.com/python-diamond/Diamond

diamond's Issues

Failed to import module: openvpn -- python 2.4 / centos 6.2 / x64

[2012-07-13 14:52:39,019] [MainThread] Failed to import module: openvpn. Traceback (most recent call last):
File "/usr/lib/python2.4/site-packages/diamond/server.py", line 156, in load_collectors
mod = import(modname, globals(), locals(), ['*'])
File "/usr/share/diamond/collectors/openvpn/openvpn.py", line 89
with open(filename, 'r') as fd:
^
SyntaxError: invalid syntax

i have default python running:
Python 2.4.3 (#1, Jun 18 2012, 08:55:23)
[GCC 4.1.2 20080704 (Red Hat 4.1.2-52)] on linux2

Disk Collector only works in python2.6

Looks like namedtuple() was used, which breaks on 2.4 installations.

[2012-01-13 09:43:19,380] [MainThread] Failed to import module: DiskSpaceCollector. Traceback (most recent call last):
File "/usr/lib/python2.4/site-packages/diamond/server.py", line 143, in load_collectors
mod = import(modname, globals(), locals(), ['*'])
File "/usr/share/diamond/collectors/Disk/DiskSpaceCollector/DiskSpaceCollector.py", line 19
name = 'root' if name == '_' else name
^
SyntaxError: invalid syntax

[2012-01-13 09:43:19,380] [MainThread] Failed to import module: disk. Traceback (most recent call last):
File "/usr/lib/python2.4/site-packages/diamond/server.py", line 143, in load_collectors
mod = import(modname, globals(), locals(), ['*'])
File "/usr/share/diamond/collectors/Disk/disk.py", line 6, in ?
from collections import namedtuple
ImportError: cannot import name namedtuple

Feature Request: over the network collector updates

Would be really fantastic to be able to push out collector scripts over the network to hosts, rather than my current setup, which is to create an rpm, and push that out.

With that feature, you could have a central console of the configs, and make it easy to push out changes, enable/disable plugins, etc.

Just a thought.

Sean

Question: Dashboard for Diamond?

Hi,

we are currently deploying diamond on some of our production machines and the tons of data gathered by diamond paired with the performance of graphite are just awesome!

Now we came to the point where we need a easy to use and in the best case, an automatically managed dashboard for the machines monitored by diamond.

Are there any ready to use implementations?

If there aren´t any, i´m willing to build something for example using gdash if there is serious interest by the community - i would create a sperate github project for this.

I just want to ask first if there are any existing implementations i missed to find.

Regards,
Christian

HAProxyCollector is missing unit test

config.py failing under python 2.4 / centos 6.2 / x64

cent02 Diamond]$ sudo python config.py
File "config.py", line 164
except IOError as (errno, strerror):
^
SyntaxError: invalid syntax

ConnTrackCollector - NameError: global name 'COMMAND' is not defined

Traceback (most recent call last):
  File "/Users/kormoc/Dropbox/Projects/Diamond/src/diamond/collector.py", line 177, in _run
    self.collect()
  File "src/collectors/ConnTrackCollector/ConnTrackCollector.py", line 26, in collect
    if not os.access(COMMAND[0], os.X_OK):
NameError: global name 'COMMAND' is not defined

Custom hostname for collectors

Hi,

Maybe I understood incorrectly how to use the custom hostname...

Or maybe it is broken?

Here how I got it to work...

pablolb@28a90e0

Best regards,
Pablo

MySQLCollector is missing unit test

Hadoop Collector issues

Per #72 (comment)

We need to change the way we build the metric keys for the Hadoop collector, as it collides on certain metrics lines in the log file.

tehmaze might be able to get a more official idea of what keys to use for the metrics.

Question: collectors config ?

I suppose that in conf/collectors were example config files, but now this directory is empty and is listed in .gitignore ... I think it's mistake.

RabbitMQCollector - raise NetworkError("Error: %s %s" % (type(out), out))

Traceback (most recent call last):
  File "/Users/kormoc/Dropbox/Projects/Diamond/src/diamond/collector.py", line 177, in _run
    self.collect()
  File "src/collectors/RabbitMQCollector/RabbitMQCollector.py", line 35, in collect
    for queue in client.get_queues():
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pyrabbit/api.py", line 420, in get_queues
    queues = self.http.do_call(path, 'GET')
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pyrabbit/http.py", line 97, in do_call
    raise NetworkError("Error: %s %s" % (type(out), out))
NetworkError: Error: <class 'socket.error'> [Errno 61] Connection refused

network.py error on CentOS 6.2 x64

new to this.. so if a newbie error please excuse.

trying to get network.py working on a centos 6.2 64bit system and see an error...

Failed to import module: network. Traceback (most recent call last):
File "/usr/lib/python2.4/site-packages/diamond/server.py", line 156, in load_collectors
mod = import(modname, globals(), locals(), ['*'])
File "/usr/share/diamond/collectors/network/network.py", line 64
greed = '0-9' if self.config['greedy'].lower == 'true' else ''
^
SyntaxError: invalid syntax

my NetworkCollector.conf is such:

cat /etc/diamond/collectors/NetworkCollector.conf
byte_unit = megabyte
interfaces = eth0
enabled = True
greedy = true

i have default python running:
Python 2.4.3 (#1, Jun 18 2012, 08:55:23)
[GCC 4.1.2 20080704 (Red Hat 4.1.2-52)] on linux2

Typo in init script causes daemon not to stop correctly on redhat systems

Therefore, restarts are broken too -- causing two instances of diamond to run.

At https://github.com/BrightcoveOS/Diamond/blob/master/dist/diamond.init.in#L77 :

-killproc -p ${DIAMODN_PIDFILE} $PROG +killproc -p ${DIAMOND_PIDFILE} $PROG

No such file or directory: '/var/log/diamond/diamond.log'

Hi,
I run make buildrpm, installed diamond on centos 6.2, copied the config and adapted it, then when trying to start it:

[root@nyvimeodfs1 ~]# /etc/init.d/diamond start
Starting diamond: Error occurs when initialize logging: [Errno 2] No such file or directory: '/var/log/diamond/diamond.log'
                                                           [  OK  ]
[root@nyvimeodfs1 ~]# less /var/log/diamond/archive.log
/var/log/diamond/archive.log: No such file or directory
[root@nyvimeodfs1 ~]# pgrep -fl diamond
[root@nyvimeodfs1 ~]# mkdir /var/log/diamond/
[root@nyvimeodfs1 ~]# /etc/init.d/diamond start
Starting diamond: WARN: Bogus pid file was found. I deleted it.
                                                           [  OK  ]
[root@nyvimeodfs1 ~]# less /var/log/diamond/archive.log
[root@nyvimeodfs1 ~]# pgrep -fl diamond
20168 /usr/bin/python /usr/bin/diamond -p /var/run/diamond.pid

so, when trying to start, it aborts because the log dir does not exist. a (temporary?) fix is to manually create the log dir as shown above.

MongoDBCollector - raise AutoReconnect(', '.join(errors))

Traceback (most recent call last):
  File "/Users/kormoc/Dropbox/Projects/Diamond/src/diamond/collector.py", line 177, in _run
    self.collect()
  File "src/collectors/MongoDBCollector/MongoDBCollector.py", line 35, in collect
    conn = pymongo.Connection(self.config['host'],read_preference=ReadPreference.SECONDARY)
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pymongo/connection.py", line 290, in __init__
    self.__find_node()
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/pymongo/connection.py", line 586, in __find_node
    raise AutoReconnect(', '.join(errors))
AutoReconnect: could not connect to localhost:27017: [Errno 61] Connection refused

UPSCollector - OSError: [Errno 2] No such file or directory

Traceback (most recent call last):
  File "/Users/kormoc/Dropbox/Projects/Diamond/src/diamond/collector.py", line 177, in _run
    self.collect()
  File "src/collectors/UPSCollector/UPSCollector.py", line 24, in collect
    p = subprocess.Popen(['/bin/upsc', self.config['ups_name']], stdout=subprocess.PIPE)
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 679, in __init__
    errread, errwrite)
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 1249, in _execute_child
    raise child_exception
OSError: [Errno 2] No such file or directory

DiskUsageCollector error

$ sudo diamond --configfile=/etc/diamond/diamond.conf --foreground -r /usr/share/diamond/collectors/DiskUsageCollector/DiskUsageCollector.py
[2012-05-28 01:20:17,585] [MainThread] Changed UID: 0 () GID: 0 ().
[2012-05-28 01:20:17,586] [MainThread] Loaded Handler: diamond.handler.GraphitePickleHandler.GraphitePickleHandler
[2012-05-28 01:20:17,587] [MainThread] GraphiteHandler: Failed to connect to brampton1-admin01.fatbox.ca:2004. [Errno -2] Name or service not known
[2012-05-28 01:20:17,587] [MainThread] Loaded Handler: diamond.handler.ArchiveHandler.ArchiveHandler
[2012-05-28 01:20:17,589] [MainThread] Loading Collectors from: /usr/share/diamond/collectors/DiskUsageCollector
[2012-05-28 01:20:17,589] [MainThread] Loaded Module: DiskUsageCollector
[2012-05-28 01:20:17,590] [MainThread] Loaded Collector: DiskUsageCollector.DiskUsageCollector
[2012-05-28 01:20:17,590] [MainThread] Initialized Collector: DiskUsageCollector
[2012-05-28 01:20:17,590] [MainThread] Scheduled task: DiskUsageCollector
[2012-05-28 01:20:17,590] [MainThread] Started task scheduler.
[2012-05-28 01:20:18,591] [Thread-1] Collecting data from: DiskUsageCollector
[2012-05-28 01:20:18,592] [MainThread] Stopping task scheduler.
[2012-05-28 01:20:18,593] [Thread-1] Traceback (most recent call last):
File "/usr/lib/pymodules/python2.6/diamond/collector.py", line 177, in _run
self.collect()
File "/usr/share/diamond/collectors/DiskUsageCollector/DiskUsageCollector.py", line 112, in collect
value = diamond.convertor.binary.convert(value = value, oldUnit = 'kB', newUnit = self.config['byte_unit'])
AttributeError: 'module' object has no attribute 'convertor'

[2012-05-28 01:20:18,593] [MainThread] Stopped task scheduler.
[2012-05-28 01:20:18,593] [MainThread] Exiting.

$ cat /etc/diamond/collectors/DiskUsageCollector.conf
enabled = True

$ dpkg -l diamond
Desired=Unknown/Install/Remove/Purge/Hold
| Status=Not/Inst/Conf-files/Unpacked/halF-conf/Half-inst/trig-aWait/Trig-pend
|/ Err?=(none)/Reinst-required (Status,Err: uppercase=bad)
||/ Name Version Description
+++-======================================-======================================-============================================================================================
ii diamond 0.2.0 System statistics collector for Graphite.

Sadness.

PowerDNSCollector is missing unit test

Is there a way to disable a collector from the config file?

I can just remove the collector ofcourse, but I wonder if there is a way to let's say, specify enabled = 1 in the config. I can build it in otherwise...

UPSCollector is missing unit test

PowerDNSCollector - OSError: [Errno 2] No such file or directory

Traceback (most recent call last):
  File "/Users/kormoc/Dropbox/Projects/Diamond/src/diamond/collector.py", line 177, in _run
    self.collect()
  File "src/collectors/PowerDNSCollector/PowerDNSCollector.py", line 27, in collect
    sp = subprocess.Popen([self.config['pdns_control'], "list"], stdout=subprocess.PIPE)
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 679, in __init__
    errread, errwrite)
  File "/opt/local/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 1249, in _execute_child
    raise child_exception
OSError: [Errno 2] No such file or directory

Invoking diamond with -r (run once) doesn't actually run once

The docs say it will run the given collector once and exit, but it lies. It runs the collector forever.

Move Handlers

Move handlers to handlers/null.py from handlers/NullHandler.py

Standardize subprocess usage in collectors

We have a bunch of collectors that all spawn off subprocesses to gather metrics. We should standardize on a way to handle the following:

• Binary path configuration, including short $PATH style usage
• Missing binaries

`make buildrpm' doesn't succeed on CentOS 6

Hey,

The new setup.py based buildrpm target doesn't quite work: https://raw.github.com/gist/4210d96d443624174c74/7e98613585e9d2789420c5581538876337e1194b/gistfile1.txt

Opening an issue here while I explore fixes.

-Naresh

Debian package doesn't build

Trying to build .deb on Debian Squeeze, here's the last lines of output where the error occurs:

dh_usrlocal: debian/diamond/usr/local/bin/diamond is not a directory
rmdir: failed to remove `debian/diamond/usr/local/bin': Directory not empty
dh_usrlocal: rmdir debian/diamond/usr/local/bin returned exit code 1
make: *** [binary] Error 1
dpkg-buildpackage: error: fakeroot debian/rules binary gave error exit status 2
debuild: fatal error at line 1325:
dpkg-buildpackage -rfakeroot -D -us -uc failed

It appears that you're not allowed to install binaries in /usr/local/bin on Debian? For now I'm not sure how to fix this...

Force collector scan/reload conf files on sighup/siguser

It would be very useful to be able to send a signal to diamond to force a collector reload/config reload on demand rather then waiting for the interval to expire.

Move Collectors

Collectors should be

src/collectors/loadavg/loadavg.py

The class name can still be LoadAverageCollector.

Feature Request: Auto disable of failed tests

How about a config parameter that can be set to a number of failures for a collector before diamond stop trying to run the collector. If I don't have mysql installed, run it X times, but then only check once per day/per hour/configurable time period.

LMSensorsCollector is missing unit test

Debian update to run as diamond user by default?

Hi all,

I haven't used Diamond extensively, and therefore haven't tried the majority of plugins, but what are your thoughts around these types of changes to the Debian scripts?

https://github.com/mattt416/Diamond/commit/3eda762387c9a918ba6fc5b2c23adece1a14666b

It'd probably be wise to ignore the local mod I made to diamond.conf.example, as that would conflict with other installation mediums.

One additional thing I noticed is the pid file is getting created w/ a perm of 666, most likely due to os.umask(0) in /usr/bin/diamond. Should the pid file be locked down a bit more than that?

Anyway, perhaps someone with some Debian packaging experience could comment?

Thanks!

Multiple Collector instances

Hi,

Today I was trying to setup Diamond to collect data from several servers, each containing multiple instances of memcached.

I cannot see how to accomplish this right now, as the collectors are added to a dict with their class name as key.

I have created a branch in my fork with a solution: https://github.com/pablolb/Diamond/tree/multiple-collector-instances.

But maybe you have a workaround or you'd prefer a better way to solve this...?

In the first commit (pablolb@f2397aa) I scan collectors config directory and schedule new instances of the collector.

In the second commit, I use MemcachedCollector as an example as how this can be used... I've seen this used in Genttoo network scripts I think (using symlinks and interpreting the actual name)

pablo@localhost collectors $ ls -l
total 4
-rw-r--r-- 1 root root 61 Mar  5 22:40 MemcachedCollector.conf
lrwxrwxrwx 1 root root 23 Mar  5 22:40 MemcachedCollector-localhost:11212.conf -> MemcachedCollector.conf

pablo@localhost collectors $ cat MemcachedCollector.conf

enabled=True
override_hostname=True
custom_name_is_host=True

RedisCollector is missing unit test

NginxCollector is missing unit test

BindCollector - TypeError: %d format: a number is required, not str

Traceback (most recent call last):
  File "/Users/kormoc/Dropbox/Projects/Diamond/src/diamond/collector.py", line 177, in _run
    self.collect()
  File "src/collectors/BindCollector/BindCollector.py", line 44, in collect
    req = urllib2.urlopen('http://%s:%d/' % (self.config['host'], self.config['port']))
TypeError: %d format: a number is required, not str

CPUCollector

Maybe I'm dense, but it seems that this collector only uploads jiffies from /proc/stat, and I can't for the life of me figure out how to get graphite to display percentages.

What am I missing?

VarnishCollector is missing unit test

Failed to import module: hadoop -- python 2.4 / centos 6.2 / x64

[2012-07-13 14:52:38,946] [MainThread] Failed to import module: hadoop. Traceback (most recent call last):
File "/usr/lib/python2.4/site-packages/diamond/server.py", line 156, in load_collectors
mod = import(modname, globals(), locals(), ['*'])
File "/usr/share/diamond/collectors/hadoop/hadoop.py", line 51
with open(filename, 'r') as fd:
^
SyntaxError: invalid syntax

i have default python running:
Python 2.4.3 (#1, Jun 18 2012, 08:55:23)
[GCC 4.1.2 20080704 (Red Hat 4.1.2-52)] on linux2

ping collector issue?

Hi.
Is it ok that ping collector generate such path [path_prefix].[host].ping.ping.[target_host]?

Stop/start script does not work properly in Ubuntu

/etc/init.d/diamond stop does not work in ubuntu

upstart is handling these things in Ubuntu. while debian require /etc/init.d/*

I can provide an upstart script for ubuntu if you want.

I did not wanted to add it myself, because I got stuck in the Ant configurations... so if you have a way to separate debian from ubuntu, the package can contains an upstart script instead of init.d

systemcollectors.CPUCollector missing newer Linux cpu metrics

CPUCollector currently collects user, nice, system, idle, iowait. It does not collect additional columns which appear with newer Linux kernels.

Here are the columns which may appear (in order):

• irq - since 2.6.0-test4
• softirq - since 2.6.0-test4
• steal - since 2.6.11
• guest - since 2.6.24

A sample line from 2.6.18-164.15.1.el5xen (only 8 columns):
cpu 54524917 1184 19017460 2207399508 25501525 0 1380737 5191044

A sample line from 3.2.0-25-generic-pae (10 columns!?!):
cpu 667245 6544 90394 33314088 111332 0 1240 0 0 0

Diamond does not automatically logrotate its logs.

Given that Diamond uses Python's logging utility, perhaps it should implement log rotation using that?

ImportError: No module named handler.Handler

Mac OS X Lion, installed Diamond from git. Running it gives this error:

$ diamond
Traceback (most recent call last):
File "/usr/bin/diamond", line 40, in
from diamond.server import Server
File "/Library/Python/2.7/site-packages/diamond/server.py", line 21, in
from diamond.handler.Handler import Handler
ImportError: No module named handler.Handler

Seems like I'm missing something?

reason for default precision to be 0?

Is there a reason for this.. I've noticed on servers with little traffic the bits are almost always 0 unless you change the precision to something like 6. It seems to break checks that might require a sub 1 metric since it will always 0 it out. I didn't see a config option for this also.

crashes when trying to run as user nobody

runs fine as root, but when trying to run as user nobody, the process stops and doesn't say why.
also, I needed to touch and chown the logfile first. see below:

[root@nyvimeostatsdev1 ~]# egrep 'user|group' /etc/diamond/diamond.conf
# Leave empty to use the current user
user = nobody
# Leave empty to use the current group
group = nobody
username    = root
[root@nyvimeostatsdev1 ~]# /etc/init.d/diamond start
Starting diamond: Error occurs when initialize logging: [Errno 13] Permission denied: '/var/log/diamond/diamond.log'
                                                           [  OK  ]
[root@nyvimeostatsdev1 ~]# ps aux | grep diamond
root      4488  0.0  0.0 103232   820 pts/0    S+   05:01   0:00 grep diamond
[root@nyvimeostatsdev1 ~]# less /var/log/diamond/diamond.log
/var/log/diamond/diamond.log: No such file or directory
[[root@nyvimeostatsdev1 ~]# touch /var/log/diamond/diamond.log
[root@nyvimeostatsdev1 ~]# chown nobody:nobody /var/log/diamond/diamond.log
[root@nyvimeostatsdev1 ~]# /etc/init.d/diamond start
Starting diamond:                                          [  OK  ]
[root@nyvimeostatsdev1 ~]# ps aux | grep diamond
root      4542  0.0  0.0 103232   820 pts/0    S+   05:01   0:00 grep diamond
[root@nyvimeostatsdev1 ~]# cat  /var/log/diamond/diamond.log
[2012-07-17 05:01:37,251] [MainThread] Changed UID: 99 (nobody) GID: 99 (nobody).
[2012-07-17 05:01:37,251] [MainThread] Detaching Process.
[root@nyvimeostatsdev1 ~]# 

SNMPInterfaceCollector inherits from unimportable class?

SNMPInterfaceCollector inherits from SNMPCollector, but I don't see how it's expected to be importable, since /usr/share/diamond/collectors (on my system) was not installed into any kind of Python-aware namespace. Browsing through some past commits, it seems that SNMPCollector used to live inside of diamond.collector, but it doesn't live there anymore.

As a result, when I start diamond, I get errors when it tries to load SNMPInterfaceCollector, saying it can't import SNMPCollector.

Debian Install has an error message

Imported from #43

Debian installs work fine, but throw this error on install. We should fix this.

vagrant@graphite:~/build/Diamond/build [master]$ sudo dpkg -i diamond_2.0.0_all.deb
Selecting previously deselected package diamond.
(Reading database ... 47779 files and directories currently installed.)
Unpacking diamond (from diamond_2.0.0_all.deb) ...
Setting up diamond (2.0.0) ...
insserv: warning: script 'K01statsd' missing LSB tags and overrides
insserv: warning: script 'statsd' missing LSB tags and overrides
Traceback (most recent call last):
  File "/usr/bin/diamond", line 18, in <module>
    from diamond.server import Server
  File "/usr/lib/pymodules/python2.6/diamond/server.py", line 21, in <module>
    from diamond.handler.Handler import Handler
ImportError: No module named handler.Handler
Processing triggers for python-support ...

PostgresqlCollector is missing unit test

Move collector tests

Move collector tests to something like "src/collectors/ExampleCollector/test/testexample.py". This is more pythonic,
and keeps tests and code separate.