[hobbit] New version of beastat.pl netapp.pl and dbcheck.pl

[Francesco Duranti]

> 
> > There's an error in the README ... 
> > If you're using hobbit 4.2 and NetApp with Ontap version 
> 6.5+ just do
> > this:
> 
> I've got it working now, and made a few tweaks so that the 
> nfs ops (the only thing we use) were tracked on their own 
> graph, instead of as a part of a group.  Makes the graph 
> scaling easier on the eyes.  ;)
>

I'll create some more definition for the next version so the user can
just choose what to use... I'm also using only NFS on some filers and
NFS/CIFS on some others or just ISCSI on another one...
 
> > For snapshot there's not a simple way, you've to edit the 
> source code 
> > of hobbit and some will get wrong. At this moment I'm finishing to 
> > work to module for rrd implementation inside the hobbit_rrd 
> for them 
> > and I've something done but I'm waiting for some nice feature 
> > regarding custom rrd implementation from Henrik so you can have 
> > something in the next few days (snapshot, quotas, qtree, 
> and something 
> > more in the next future like full nfsstatistics and 
> interface/network statistics)...
> 
> I was going to just hack up the do_disk.c file and see what I 
> could do, but sounds like waiting is the better option.  
> filerstats2bb used to put the .snapshots filesystem alongside 
> those in the disk column, which worked out pretty well except 
> for segregating the snapshot alerts to separate users.

I've preferred to put the snapshot out of disk space report to have a
clean situation ... 
A warning on the snapshot can be a problem related to an oracle
snapshot/backup that didn't worked so I've to send alert to db admin to
check them... 
If you want to give it a try I can send a patch working with 4.2.0 (it
should work from 4.2.0 to the last snapshot) but this will include also
the graphs rrd function for dbcheck.pl and beastat.pl script and is a
work in progress so it could work but it could also do bad thing :P
(well it's running now by one week on my test and production system
without problems).

> 
> > If you want data to be reported silently without having a 
> column just 
> > to create a graph in the trend you can also do:
> > bbopscolumn             = ops:data
> > bbstatscolumn           = stats:data
> 
> Again, very cool. I now have an ops column on my front page, 
> displaying NFS data, and it is a welcome site.  :)
> 
> I was going to try and get a way to collect the disk 
> %utilization values for my volumes (those are displayed when 
> you collect data with statit for a period of time).  That 
> would also be a extremely useful data point.

I'm trying to define what data to get next... The ifinfo also seems
interesting (it also seems the only method to get byte counter related
to network interface). I think the next think I'll get will be: cluster
status, nfsstat, graphs for snapmirrored volumes and data related to
snapvaulted volumes. 
I've saw the statit command and it's nice the only thing not so good is
the fact that it work in background and there can be some problem
getting good data out of it because running wrong command could get
wrong data on that. I was also checking the "stats show" command that
seems to report the same kind of data and can be run like the sysstat so
with a 300 seconds interval that will let me get the data at the next
run of the script. I should try to understand what of all those data
will really be usefull to monitor and how to graph them .... 


Francesco