RE: [hobbit] Xymon 4.3.0: Beta version available on Sourceforge

["Everett, Vernon" <Vernon.Everett (at) woodside.com.au>]

Your comment reminds me of this
http://www.flickr.com/photos/straup/2247714432/

On a more serious note, I look forward to storing hobbit information in a relational database.
Imagine the reporting flexibility we could have with SQL.
- Give me all server with root file system smaller than 8gb that have been over 80% in the last 3 months.
- give me a list of all servers with operating system X that are running custom test Y.
There will be no limits to the level of reporting.

I think it would be a bad idea to abandon the old RRD in favour of RDB, but to rather have the ability to use RRD by default, and RDB as an additional reporting structure.
This would limit the dependency issue.

Using an RDB would also allow for custom defined data points.
Currently, with RRD, we only store 48 hours of data points at 5 minutes granularity.
In an RDB we could define our own retention period.
Because of the way RDBs work, there would probably need to be a purge job to purge old data, and accumulate it into data points for the table storing the next level of granularity.

It could be an interesting project, but would need to be very carefully constructed to prevent performance issues.
Do we have any database architects on the list?

Regards
     Vernon


________________________________
From: Brian Catlin [mailto:bcatlin (at) gmail.com]
Sent: Tuesday, 10 February 2009 11:43 AM
To: hobbit (at) hswn.dk
Subject: Re: [hobbit] Xymon 4.3.0: Beta version available on Sourceforge

How about you alert on the incoming as usual, then store the events (batched or not) into the backend database so that reports can be developed for management, history could use for lookups, etc... That might make this a little more palatable to you, and a lot more usable to those of us who have to prove to management that the  time we invest in getting this doing what we want is not a drain on their resources.

Reports are management blood, no reports - no use... (for managers) - I have had managers tell me this in the past. (on the other hand - two weeks after I installed clients on one department's sun servers, the manager had some questions about server performance over the past few weeks = was quite pleased and surprised when I gave him a spreadsheet that laid out the data for all his servers in 1/2 hour, as was even more please when I showed him how to view his page of servers.)

Another thing that using a database to store history in - much better security for those of us who work in security realms and have to protect data - alarm history can give a lot of info away to a hacker if they gain access to the server running the master.
With a database, it can be secured and only have authorized users that access it.

One thing everyone should remember - in business, as in life, perception is 80 % of the way things are seen - think about it...


lurch (at) inorbit.com
-------Original Message-------

From: Henrik Størner<mailto:henrik (at) hswn.dk>
Date: 2/9/2009 4:40:49 PM
To: hobbit (at) hswn.dk<mailto:hobbit (at) hswn.dk>
Subject: Re: [hobbit] Xymon 4.3.0: Beta version available on Sourceforge

Hi Giovanni,

On Mon, Feb 09, 2009 at 07:21:43PM -0200, giovanni (at) redix.com.br<mailto:giovanni (at) redix.com.br> wrote:
>
> Hi Henrik,
>
>    I was reading about 4.5 release and maybe its time to think about store
> the data into a Database like MySQL/PostgreSQL, if you do that far more
> contributors can start to colaborate with customs web-interfaces, reports,
> etc etc.

this subject comes up once in a while. I have two problems with it:


1) I've never done any programming that interfaces to a database, so
it would mean digging into one more API - but that is something I
could probably handle.

2) The real problem is that unless the database is co-located with the
Xymon server, then your monitoring suddenly becomes dependant on
a remote database-server. So what happens if your database server loses
the network connection to the Xymon server ? You won't get any alerts
from Xymon, because it has no data available to generate alerts from.


It's the same reason that has kept me from using SAN storage for my
production Xymon server at work. Much too complex for my taste, far
too many ways it can break.

And the time when you need your Xymon server the most is *exactly*
when everything else has gone down in flames.


What I *would* consider is to create a module like the hobbitd_filestore
module, except it sends the status-data off to a database somewhere.
Or even an external module for hobbitd_rrd that gets all of the
parsed data we collect and use as the basis of all of the graphs.
Such modules would be very easy to write for someone who knows how
to do programming with the database API - my guess is that it wouldn't
take more than a day or two.

So if you know someone who could voluteer for such an add-on, I would
be very happy to work together with him/her on putting such an add-on
together.


Regards,
Henrik


To unsubscribe from the hobbit list, send an e-mail to
hobbit-unsubscribe (at) hswn.dk<mailto:hobbit-unsubscribe (at) hswn.dk>






NOTICE: This email and any attachments are confidential. 
They may contain legally privileged information or 
copyright material. You must not read, copy, use or 
disclose them without authorisation. If you are not an 
intended recipient, please contact us at once by return 
email and then delete both messages and all attachments.