The system statistics collection daemon "collectd"

teichm | 17 Jun 2013 15:54

Assertion crash in collectd 5.2.1 / libgcrypt 1.5.2

Heyho,

On our collectd 5.2.1 from the Debian Package 5.2.1-1 we encountered an 
assertion crash from libgcrypt11 1.5.2-1 that occurs aproximately once a 
day on random conditions.

I do not know, if collectd is using libgcrypt wrong or if it is a bug 
in libgcrypt.
It seems, ath_install was called with the first argument beeing NULL. 
At least i have no other explanation of why this codepath would be 
reachable.

Since i am also working on my Bachelor's thesis at the moment in which 
I am developing Monkey, a new automated bug reporting system, i 
generated a report for your convenience:
https://gnunet.org/monkey/examples/collectd-01/scrutinizer.html

If you are interested in the Debugging System, you can find more 
information on:
https://gnunet.org/monkey/

Also i have a survey (or rather ‘experiment’) running, in which you 
will get three different crash reports (stack trace, stack trace + *life 
gdb session* on a core dump and of course one generated from Monkey) and 
try to find the respective bug.
You can participate here:
https://gnunet.org/monkey/survey/index.php/943594?lang=en

--Markus

(Continue reading)

headers

teichm | 18 Jun 2013 08:38

Assertion crash in collectd 5.2.1 / libgcrypt 1.5.2

Heyho,

On our collectd 5.2.1 from the Debian Package 5.2.1-1 we encountered an 
assertion crash from libgcrypt11 1.5.2-1 that occurs aproximately once a 
day on random conditions.

I do not know, if collectd is using libgcrypt wrong or if it is a bug 
in libgcrypt.
It seems, ath_install was called with the first argument beeing NULL. 
At least i have no other explanation of why this codepath would be 
reachable.

Since i am also working on my Bachelor's thesis at the moment in which 
I am developing Monkey, a new automated bug reporting system, i 
generated a report for your convenience:
https://gnunet.org/monkey/examples/collectd-01/scrutinizer.html

If you are interested in the Debugging System, you can find more 
information on:
https://gnunet.org/monkey/

Also i have a survey (or rather ‘experiment’) running, in which you 
will get three different crash reports (stack trace, stack trace + *life 
gdb session* on a core dump and of course one generated from Monkey) and 
try to find the respective bug.
You can participate here:
https://gnunet.org/monkey/survey/index.php/943594?lang=en

--Markus

(Continue reading)

headers

Brendan Horan | 12 Jun 2013 15:02

Collectd, snmp linux interface stats

Hi List,

I am really enjoying collectd, thank you.

I have two issues with collectd and SNMP plugin.

I am using the following block to collect interface traffic stats :
   <Data "std_traffic">
       Type "if_octets"
       Table true
       Instance "IF-MIB::ifDescr"
       Values "IF-MIB::ifInOctets" "IF-MIB::ifOutOctets"
   </Data>

This works fine for my Cisco kit, but all my Linux hosts never get any data inside collectd.
That seems odd becase if I run the following  "snmpwalk -v2c -c public $HOSTNAME IF-MIB::ifInOctets"

I get back the expected values.

I don't understand what I am doing wrong. The above works for my Cisco kit, works via snmpwalk.

The second issue I have and of much less importance is that on my Cisco switch the above data block only brings
back one VLAN.
On my Cisco Router all VLAN's are brought back.

Unsure why my switch won't get all the VLANS, all physical switchports are fine.

Can someone give me a hand ?

Thanks

(Continue reading)

headers

Беркович Данил | 6 Jun 2013 09:33

collecd wishes

Good day!

In collected metrics I hadn't found two very interesting for me.
I don't found metric "disk queue length" for disk plugin and "media connected" or so on for netlink or interface plugin
I think it will be helpful

Thanks forward!
Danil

--
С уважением
ведущий специалист отдела системного администрирования
службы развития и эксплуатации OSS систем
департамента ИТ ОАО "Мегафон", Кавказский филиал, г. Краснодар
Беркович Данил
мобильный: +7(928)201-12-60
служебный: (861)291-00-42
внутренний: 1842
mailto: danil.berkovich-uHXVZI0VGm4vJsYlp49lxw@public.gmane.org

_______________________________________________
collectd mailing list
collectd@...
http://mailman.verplant.org/listinfo/collectd

headers

William Salt | 17 Jun 2013 02:00

processes plugin process by name - percent of all cpus feature

Im using the processes plugin to find specific processes via regex, and monitoring their cpu usage, this is the total time spent in jiffies (milliseconds) across all cores.

In graphite, i am dividing it by the number of cpu cores i have on a machine then turning that into percent. this gives me a percentage of total cpu usage across all cores for that process.

However, i cannot ascertain in graphite the exact number of cpus a node has, it differs greatly, and its not scalable to have static values in graph query strings like this, nor efficient to work this out when rendering a graph.

I wondered if someone could patch the plugin to allow a flag to produce the average cpu use across all cores? grabbing the number of cpus from /proc/cpuinfo and dividing the metric by that? bonus if it had a flag to output a percent!

Cheers

Will

_______________________________________________
collectd mailing list
collectd@...
http://mailman.verplant.org/listinfo/collectd

headers

David Halko | 14 Jun 2013 18:25

CollectD, Collection4 (C4), and General Questions

Hello List,

I have been following this software for years, did some basic
configuraiton in the past, but I am thinking about getting serious
about it.

The multicast send from a single collector to multiple receivers is
one of the most innovative features I have seen!
It is great to see that thresholding able to be sent to syslog - which
means collectd can be integrated with simplicity (really wish SNMP
Traps were supported.)

I have a couple of key questions concerning CollectD (SNMP plug-is
mostly required)
- I remember there being a top-end number of devices that a CollectD
instance could collect against... (Was it open file handles? Can
CollectD go up to 4096 yet?)
- Is there any known issues with SNMPv3 support?
- Can strings be collected daily (i.e. firmware, sysdescr, interface
names, etc.) and stored so we can check for changes?
- Is thresholding on string changes be done? (i.e. firmware update
noted from a string value change from last poll, notification that a
reboot caused interfaces to scramble from last poll, etc?)
- Is thresholding on sysuptime reset available? (i.e. sysuptime ticks
becomes lower than it was before, but some type of logic check to
ignore flip over?)

I have some general question concerning C4 and usability.
There is not much activity (updates) on this package, nor much
advertizing of C4 on the main CollectD page, but it really seems to
fit a lot of needs.
- is C4 stable for produciton work?
- are tabular reports possible (i.e. showing interfaces on a device,
firmwares across many devices, last reboot time across many devices,
etc?)
- if collecting on databases and routers, is there a way to ensure the
two groups are isolated from one another when they try to view the
RRD's?
- if collecting on databases and routers, where the two groups are
isolated from each other, is there a single-view where the admin (i.e.
me) can see it all?

My need is not so much fast collecting, but I have some HORRIBLY
CHEAP, SLOW, HIGH LATENCY, and UNRELIABLE cellular devices (~2500) and
I would like to start assembling some graphs (low bandwidth, so
occasional SNMPv1 collections ~15 minute intervals.) I also have a few
servers & databases on those servers I would like to SNMPv3 poll.
Obviously, the two groups are uninterested in each other's data and I
have a need to enforce separation.

I would appreciate some thoughts!

Thanks!

headers

Michael Sioutis | 10 Jun 2013 16:04

Compiling collectd with lxc

Dear developers,

I am trying to compile collectd with lxc enabled.

I have cloned the following repo in a 64bit Fedora 12 (Constantine) machine:

https://github.com/dotcloud/collectd/tree/lxc-plugin-package

After that, I installed some required dependencies:

...

flex

byacc

autoconf

automake

libtool

pkgconfig

libtool-ltdl

libtdb

libtool-ltdl-devel

libgcrypt-devel

perl-ExtUtils-CBuilder

perl-ExtUtils-MakeMaker

perl-ExtUtils-Embed

libvirt-devel

lxc-devel

libcap-devel

...

I issued the ./build.sh script which ran flawlessly and created among others a 'configure' file.

So, as a next step, I ran the following command:

# ./configure --enable-lxc > configure.log 2>&1

This gave me a dependency error. If you like (and the list allows attachments) I can send you the whole configuration log file,

but the main issues are the following:

]# grep lxc configure.log

checking lxcstats.h usability... no

checking lxcstats.h presence... no

checking for lxcstats.h... no

configure: running /bin/sh ./configure --disable-option-checking '--prefix=/opt/collectd' '--enable-lxc' --cache-file=/dev/null --srcdir=.

liblxcstats . . . . . no (lxcstats.h not found)

lxc . . . . . . . . . no (dependency error)

So, it seems that something is missing concerning lxc, namely lxcstats libraries, but I do not know where to find those files.

Further, even if I try to enable the libvirt plugin that meets all dependencies, I still get a dependency error:

# ./configure --enable-libvirt > configure.log 2>&1

# grep libvirt configure.log

checking libvirt/libvirt.h usability... yes

checking libvirt/libvirt.h presence... yes

checking for libvirt/libvirt.h... yes

configure: running /bin/sh ./configure --disable-option-checking '--prefix=/opt/collectd' '--enable-libvirt' --cache-file=/dev/null --srcdir=.

libvirt . . . . . . . yes

libvirt . . . . . . . no (dependency error)

(Everything is 'yes', but I get a dependency error in the end...)

Any guidance and help would be much appreciated.

Best regards,

Mike

_______________________________________________
collectd mailing list
collectd@...
http://mailman.verplant.org/listinfo/collectd

headers

Wawrzek Niewodniczanski | 6 Jun 2013 18:05

Caching results and sending different results using different write mechanism

Hi,

I remember reading somewhere that, with collectd, you can "cache"
results on a client an push them in convenient time.
Is it possible?
Especially is it possible with graphite plugin?

Another question.
I see in the Chains documentation that you can filter message to send
them to different write plugins. Do you possible have examples of such
configuration?

Thanks,
Wawrzek
--
Dr  Wawrzyniec Niewodniczański    or Wawrzek for short
  PhD in Quantum Chemistry  & MSc in Molecular Engineering
   WWW: http://wawrzek.name E-MAIL: jobs <at> wawrzek.name
      Linux User #177124

_______________________________________________
collectd mailing list
collectd <at> verplant.org
http://mailman.verplant.org/listinfo/collectd

headers

Joaquin Cuenca Abela | 29 May 2013 16:46

Problem with email notifications

Hi,

I enabled the email notifications in collectd, and installed postfix in a ubuntu 12.04 to serve the emails.

As soon as I installed it, I got the same notification ~100 times, along with a few emails which seemed to be malformed (contained some chinese when viewed in gmail). See attached screenshot with the emails.

Has anybody seen something like this before? Any ideas to fix it?

--
Joaquin Cuenca Abela

_______________________________________________
collectd mailing list
collectd@...
http://mailman.verplant.org/listinfo/collectd

headers

Vedran Bartonicek | 31 May 2013 10:08

Plugin memory: new options

Hi,

It seemed useful if there would be an option to gather available memory in more detail.

Like "free" column from "free -m " in row "-/+ buffers/cache"

So there could be an option like e.g. "ReportBuffersCache", which would put to use gathering

of total used and total free memory.

Use case for that is that collectd would be able to fire a notification if available memory is below some threshold.

There could also be another option like "ReportBuffersCachePercentage" which would to the same as above but in percentages. The benefit for percentages is in e.g. cluster machines, where machines might have different RAM sizes, and is impractical to manually prepare thresholds for each machine.

I think that I will probably need those things above in my project, so if this looks reasonable, I could make a pull request.

-Vedran

P.S.

I am not sure if some other plugin is available for this already, or there is another way to do the same.

_______________________________________________
collectd mailing list
collectd@...
http://mailman.verplant.org/listinfo/collectd

headers

Chris Kittlitz | 29 May 2013 19:08

Making collection3 compatible with cgi-bin

Hello.

I had a problem recently where the graphs were loading fine on my development machine, but not on my production machine.

I've narrowed down the problem to the fact that web servers with a cgi-bin (or cgi) directory will assume everything within that directory is executable. Attempts made to access files in that directory from html will result in those files being executed instead of read. For example, if style.css is in the cgi-bin directory, then it will not be correctly read by the web browser.

The graphing tools in collection3 are all within a single collection3 directory;

$ pwd;ls
/usr/share/doc/collectd-4.9.1/contrib/collection3
bin etc lib README share

And the index.cgi and htmlcss.cgi reference the css and js files in the share directory using a relative path. For example "../share/navigate.js".

When I tried to deploy this on a tomcat server, I was forced to put this in the cgi directory as follows.

$ pwd
/usr/local/apache-tomcat-7.0.18/webapps/system/WEB-INF/cgi
$ ln -s /usr/share/doc/collectd-4.9.1/contrib/collection3 collectd

However, when I went to access the graphs via firefox, I noticed that the buttons didn't work. This was because navigate.js couldn't be loaded, so all the functions defined in it were undefined.

I found reference to this type of problem online here:
http://www.theadminzone.com/forums/showthread.php?t=6156

and here:
http://forums.jaguarpc.com/open-discussion-chit-chat/13449-external-javascript-file-problem.html

To fix this I did the following.

1. Create another link to collection3 in the webapps directory.

$ pwd
/usr/local/apache-tomcat-7.0.18/webapps
$ ln -s /usr/share/doc/collectd-4.9.1/contrib/collection3 collectd

2. Edit the index.cgi and htmlcss.cgi files to reference "/collectd/share/..." instead of "../share/...".

The graphs then load normally.

While this works, I dislike having the change the source code because it is something that will have to be repeated every time we update collectd.

Has anyone else had this problem? Have you been able to solve it better than I?

Thanks!!

Chris

_______________________________________________
collectd mailing list
collectd@...
http://mailman.verplant.org/listinfo/collectd

18	June 2013
49	May 2013
27	April 2013
66	March 2013
91	February 2013
63	January 2013
24	December 2012
59	November 2012
56	October 2012
67	September 2012
43	August 2012
55	July 2012
43	June 2012
62	May 2012
66	April 2012
60	March 2012
69	February 2012
54	January 2012
36	December 2011
27	November 2011
34	October 2011
45	September 2011
67	August 2011
47	July 2011
63	June 2011
42	May 2011
43	April 2011
87	March 2011
21	February 2011
23	January 2011
38	December 2010
44	November 2010
41	October 2010
98	September 2010
81	August 2010
46	July 2010
116	June 2010
64	May 2010
62	April 2010
104	March 2010
93	February 2010
66	January 2010
72	December 2009
44	November 2009
127	October 2009
117	September 2009
151	August 2009
111	July 2009
49	June 2009
107	May 2009
116	April 2009
103	March 2009
65	February 2009
78	January 2009
32	December 2008
39	November 2008
76	October 2008
56	September 2008
135	August 2008
49	July 2008
84	June 2008
51	May 2008
76	April 2008
119	March 2008
98	February 2008
58	January 2008
70	December 2007
104	November 2007
60	October 2007
83	September 2007
52	August 2007
17	July 2007
43	June 2007
82	May 2007
54	April 2007
30	March 2007
34	February 2007
25	January 2007
67	December 2006
145	November 2006
72	October 2006
15	September 2006
24	August 2006
34	July 2006
35	June 2006
44	May 2006
65	April 2006
43	March 2006
46	February 2006
22	January 2006
73	December 2005
71	November 2005
32	October 2005

Mon	Tue	Wed	Thu	Fri	Sat	Sun

					1	2
3	4	5	6	7	8	9
10	11	12	13	14	15	16
17	18	19	20	21	22	23
24	25	26	27	28	29	30