[Orca-users] Orcallator: Running orca on each machine

[Kalpesh Patel]

/var/orca/orcallator/machX refers to the directory where Orcallator collects
raw stats at given rate (five minutes being default) for a particular node
X. Running Orca with -once command line option produces rrd files in
/var/orca/rrd/orcallator/o_machX directory for node X. Orca will then use
rrd files created in /var/orca/rrd/orcallator/o_machX directory for node X
to generate pngs for node X. Orca with -once command line can run anywhere
where Orca is install but if you want to have an index page in hyper linked
fashion to all nodes then you need to make raw data from all nodes available
to Orca in one single place.

I am not sure if Orca is capable of generating html pages without pngs and
have Orca stitch together pngs with html pages. I do not see a way for Orca
to link back the pngs that are generated by a different process back to html
pages. Orca process is in charge of manipulating original raw rrd into its
own, generate pngs from its own rrds, and then generating html which tie
together pngs in a browse able pages that can be served by a web server with
properly constructed hyper links.

In a central place, you will need to get files from the actual node by some
means into the central place. These where you imagination can go crazy and
your SA skills kick in. Take you pick from NFS, RSYNC, RCP, RSYNC, FTP, etc.

Thus for:

> Mkdir -p /var/orca/orcallator/mach1
> Mkdir -p /var/orca/orcallator/mach2
> Mkdir -p /var/orca/orcallator/mach3

You will have to go to node mach1 in /var/orca/orcallator/mach1 directory
and copy contents of that directory over to central place in to
/var/orca/orcallator/mach1 directory. Then on to node mach2 and copy over,
etc. Once you have all of the raw rrd in one place, you run Orca with -once
option in a cron to have pngs and html generated. This is the by far the
easiest way to set it up although you will require good sized box.

If do not want to have all of them aggregated to central place then you can
have each node generate pngs and html pages. Then copy pngs and htmls over
to your web server. But before you do that, you will have to make sure that
the orcallator.cfg on all nodes which are generating pngs and htmls are
consistent other wise you will have each node's pngs and htmls stepping on
each other's with file names. Doing it this will require very intricately
choreographed dance between all the nodes.


-----Original Message-----
From: Yves Dorfsman [mailto:yves at zioup.com] 
Sent: Tuesday, February 24, 2004 6:16 PM
To: orca-users at orcaware.com
Subject: RE: [Orca-users] Orcallator: Running orca on each machine


On Sun, 22 Feb 2004, Kalpesh Patel wrote:

> All you need are the rrd files from the machines from which you want to
> plot. In particular you need to look at the orcallator.cfg file on the
> machine where you are creating html pages. In particular, interest are the
> parameters named 'html_dir', which is where the index.html will be
created,
> and 'find_files' parameter in the 'group' configuration.
>
> The way I do is that under /var/orca/orcallator directory, I create a
> sub-directory for each machine that I want the plots to be generated. Then
> in each of those sub directory, I get rrd files from that node into the
> sub-directory created in the previous step. Once all of them are copied
> over, run orca with '-once' option. Here are the steps on Solaris machine
> for my configuration:
>
> Mkdir -p /var/orca/orcallator/mach1
> Mkdir -p /var/orca/orcallator/mach2
> Mkdir -p /var/orca/orcallator/mach3

But there is no rrd files in those directories, this is were orcallator
puts its raw data. Are those the directory you are copying, or is that a
typo ?

> 1	html_dir                /tmp
> 2	find_files
> /var/orca/orcallator/(.*)/(?:(?:orcallator)|(?:percol))-
> \d{4}-\d{2}-\d{2}(?:-\d{3,})?(?:\.(?:Z|gz|bz2))?
> 3	column_description      first_line
>
> I also included third line knowing that rrd files contain first line as
the
> column description for all of the machines. This will generate an index
page
> with three machines plots.

I am looking at this, to see what I can do with it - I'll get back to the
list if I figure something out.

For information, what I am trying to do is to have each server/workstation
running orca to generate the rrd and png files, copy those on one web
server, and have orca generates the html files only on this box, without
wasting cpu time trying to re-generate the rrd and png files.


Yves.
----
                                         http://www.cuug.ab.ca/~dorfsmay
                                         http://www.SollerS.ca