For reasons that are not clear to me the EPEL build also uses a slightly
different (no libganglia?) set of packages, which can make switching
between EPEL and other builds difficult.
On 03/28/2016 12:36 PM, Adrian Sevcenco wrote:
> On 03/28/2016 06:45 PM, Damir Krstic wrote:
>> Does anyone have
Hi Javier, your issue sounds at least somewhat similar to:
https://github.com/ganglia/monitor-core/issues/47
Which includes several cross referenced discussions, but no clear fixes.
Are your clusters all on the same local network?
On 02/15/2016 12:38 PM, Javier Villar Fernández wrote:
> Hi
While there are certainly a lot of changes since 3.1.x, the architecture
descirption in 'Monitoring with Ganglia' is still quite relevant.
On 03/04/2015 02:43 PM, Ralph Castain wrote:
Hi folks
I’m looking at Ganglia for one of my projects, and see that the architecture
has evolved
I don't see anything obviously wrong with the snippet.
You can run procstat.py from the command line (--help for options) and
pass along the configuration over the command line. That might provide
a faster feedback cycle for debugging.
On 02/24/2015 03:17 PM, Grigory Shamov wrote:
Hi All,
I would be interested in mentoring this summer.
--
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing,
gmond only keeps the latest value, so the short answer is 'nothing'.
It would be a significant (but very useful) architectural change for the
various components keep a sliding window of values and take a lot of the
luck out of getting everything to line up.
On 02/12/2015 05:38 PM, Brad Hough
I'm trying to write a gmond python module that needs to measure values
greater than 2^32 (bytes of memory/storage). I'm having trouble getting
that to work as there either isn't a uint64 type or the python module is
turning all ints into 32bit [1].
What's the right way to pass 64 bit numbers
On 2014-02-18 09:39, Rushton Martin wrote:
Could you simply divide the value by 2^20 (or 10^6) and send MiB (or
MB)?
This is actually similar to what the built in memory metrics do. They
report KiB and then the fancy report converts that to something nice.
The work around that I am currently
On 08/06/2013 11:42 AM, m...@termit.ln.ua wrote:
1. Hostnames. Looks like ganglia takes hostnames from PTR records. It
looks not good for me to amend PTR record if i want to change hostname.
Is it possible to explicitly set hostnames in gmond.conf?
If memory servers I think you want to look
We have been slowly been trying to squash stability issues with gmetad.
Many of the symptoms seem to relate to sockets ending up in
CLOSE_WAIT, although I'm unsure if that is a useful clue or a side
effect. The user visible problem is metrics not getting updated and TN
climbing for an
You should see a little Trend button when clicking around in the UI
next to CSV/JSON/Inspect/Timeshift etc.
On 05/20/2013 09:14 AM, Емельянов Борис wrote:
Good day!
I want to use trendline for some of my metrics, but the only way i
manage to switch it on is to send ?trend=1trendhistory=1
You are correct that there is no way to 'group-by' arbitrary attributes.
Some process will have to pass in a set of hostnames (or regex). There
are some short examples in the views document
http://sourceforge.net/apps/trac/ganglia/wiki/ganglia-web-2#Views.
On 03/19/2013 05:32 AM, Torstensen
On 2013-03-14 09:04, Simon Boulet wrote:
Now, I would like to group VMs separately from the Hosts in Ganglia Web.
Currently both the VMs and the Hosts appears in the same cluster. I was
thinking of having a separate cluster for the VMs. However, it doesn't
seems possible in my current
I gather from the context that you are identifying hosts by ip? In that
case you could (in principle) 'mv old-name new-name' for all rrd files.
On 2013-03-17 14:39, dan.fra...@pnc.com wrote:
Due to issues beyond my control, the Linux VMWare server that runs our
top-level gmetad and apache web
Could you elaborate on what you mean by head node? Do you mean a gmond
aggregator?
How was the script failing?
On 09/25/2012 04:50 AM, deep desai wrote:
hi,
I have written a python module for getting the stats for cassandra
using the nodetool cfstats. The script is working fine for all
What makes 60 an unlucky number?
On 10/25/2012 05:20 PM, Vladimir Vuksan wrote:
60 seconds is likely the problem. I would leave it at default ie 15. I can
explain later.
Potter,Mark L mlpot...@mdanderson.org wrote:
Nicholas,
I have it set to collect every 60 seconds at the moment as
can't be 100% sure that this patch will fix your problem but it would be
worth a try.
Regards,
Nick
[1] https://github.com/ganglia/monitor-core/pull/50
On Sat, Sep 15, 2012 at 12:16 AM, Chris Burroughs chris.burrou...@gmail.com
wrote:
We use ganglia to monitor 500 hosts in multiple
Is there any known java library (or antlr file or other lexer/parser)
for libconfuse based configuration files? I'd like my java applications
to just read gmond.conf (like gmetric does) instead of having to
configure the same thing twice. I suppose a something that is custom to
ganglia would
We have all of our metrics of interest generated by Coda Hale's metrics
package: http://metrics.codahale.com/
That includes a ganglia reporter that can be used to send metrics to
gmond. (But not arbitrary pre-existing beans.)
On 09/13/2012 08:43 AM, Martin Knoblauch wrote:
Hi,
as part of a
problem are also welcome!)
Thank you,
Chris Burroughs
[1] https://github.com/ganglia/monitor-core/issues/47
[2]
120827 89
120828 6
120829 3
120830 4
120831 5
120901 1
120902 6
120903 2
120904 9
120905 4
120906 70
120907 523
120908 85
120909 4
120910 6
120911 2
120912 5
120913 5
Surge [1] is scalability focused conference in late September hosted in
Baltimore. It's a pretty cool conference with a good mix of
operationally minded people interested in scalability, distributed
systems, systems level performance and good stuff like that. You should
go! [2]
This year there
again
I found
http://bugzilla.ganglia.info/cgi-bin/bugzilla/show_bug.cgi?id=189 but
that seems to refer to an un-escaped name, as opposed to a transient but
not recovered error.
On 08/13/2012 01:29 PM, Chris Burroughs wrote:
So for background, my original problem is that load_one
On 08/15/2012 09:13 AM, Vladimir Vuksan wrote:
2 seconds seems a bit excessive. Last time I tested downloading from
gmetad on a setup with ~50k metrics it took between 250-300ms to
download XML. I will recheck later today. Can you tell me what the size
of the downloaded XML is Bytes ie.
On 08/14/2012 02:00 PM, Douglas Wagner wrote:
I don't mean to ask stupid questions, maybe I'm not reading you right...
Are you saying that it's taking 2K ms (20s) between received data sets?
(i.e. Every 2K ms you see XML data come into your application)? Or that
it's taking 2K ms between the
So for background, my original problem is that load_one will not be
updated by gmetad for a period of over 600 seconds (an arbitrary timeout
signifying that gmond/the host is probably down). It occurs a few
times/day across hundreds of hosts, and often occurs near midnight
localtime. This
I have a process that periodically polls gmetad (builds models of some
metrics, alerts if things don't look like). To reduce the number of
variables I set up a dedicated gmetad on the same host as the poller and
set write_rrds off. Unless I'm missing something the only thing it
should be doing
I'm trying to find some api for getting (not setting) events. I've
looked through events.php [1] and see a human readable output. But I'm
looking for something for the machines. Am I missing something obvious?
[1] https://github.com/ganglia/ganglia-web/blob/3.5.2/events.php
Really exciting. But I'm confused how this works with the round robin
nature of RRD. Don't we by default only have (for example) daily data
for past 24 hour period, not 48 hours?
On 05/16/2012 07:54 PM, Vladimir Vuksan wrote:
There is a blog post about a new feature in Ganglia Web called
I've also observed this and have been unable to find a solution. In my
case at least there was no obvious correlation with the number of
metrics or weather the gmond was an aggregating or not (so several
orders of magnitude in the number of metrics did not matter, it might
happen on 2 out of 80
Currently tcpconn.py uses netstat to get it's socket stats. This gives
lots of detail but is far too slow for much production use (running
netstat can take many minutes). /proc/net/sockstat gives less
information but has no performance problems. There was a suggestion
previously to use the ss
On 12/09/2011 07:51 PM, Matt Massie wrote:
What are the things you would be most interested in? Are there other
topics you'd like to see covered?
I would also like to see more details on how and why for different
common variations. For example, some people set 'host' to something
other than
On 10/27/2011 08:36 PM, Rick Cobb wrote:
In your case, your sender doesn't know about the other receivers and will
need to be configured via its own configuration technique. It can be
confusing that gmond is a daemon with three roles (and gmetad one with two),
but that's the way it is.
I have a unicast setup where each cluster box is sending to 3 collector
servers. For all sorts of normal metrics (cpu load) etc this works
fine. Each cluster box also has several java applications [1] are also
sending metrics to the gmond on localhost. Running with debug=2, I see
that those
We are trying to determine the appropriate hardware for our ganlgia web
server holding the RRD files. From what I have been able to tell most
of the tuning advice [1] [2] seems to be focused on flushing to disk
less frequently (with rrdcached, tempfs, page cache settings, etc.) and
not on tuning
On 09/20/2011 09:47 AM, Jesse Becker wrote:
I'd also suggest reading this paper:
http://www.usenix.org/event/lisa07/tech/full_papers/plonka/plonka_html/
It's about scaling MRTG up to about 320,000 RRD files (without SSDs).
Thanks Jesse I saw that paper and found it very helpful. But
So my understanding is that gmond on each node receives UDP messages
from localhost is polled by gmetad via tcp/xml. The wikipedia page [1]
states that gmond also sends data via Unicasting or Multicasting host
state in external data representation (XDR) format using UDP messages.
But I'm having
36 matches
Mail list logo
