I was running a Hobbit server on the beta package with the allinone patch July 04 2006. However, we now want the hosts being monitored by this ("old") Hobbit server accessible from our other ("new") Hobbit server (running 4.1.2p1). We can tolerate differences in log file/port/proc features for a while until we upgrade the "new" server.
So, as when migrating from BB to Hobbit, I set hobbitd to listed on port 1985 on the "old" server, set up bbproxy to forward to localhost:1985 and the "new" Hobbit server, and transferred the relevant data.
However, now the "old" server is not getting data from clients, but the "new" one is, eg cpu, disk, files, memory, msgs, ports, procs are purple on the "old" server but green on the "new" sever (and collecting in the RRDs).
The bbproxy section on the "old" server has:
[bbproxy] #DISABLED ENVFILE /usr/lib/hobbit/server/etc/hobbitserver.cfg CMD $BBHOME/bin/bbproxy --hobbitd --bbdisplay=127.0.0.1:1985,10.10.9.32:1984 --report=$MACHINE.bbproxy --pidfile=$BBSERVERLOGS/bbproxy.pid LOGFILE $BBSERVERLOGS/bbproxy.log
Any ideas, or is this a potential, or known, bug in 4.2 ? (I'm building the rc as I type ... but would appreciate any ideas).
Regards, Buchan
-- Buchan Milne ISP Systems Specialist B.Eng,RHCE(803004789010797),LPIC-2(LPI000074592)
On Fri, Jul 14, 2006 at 05:29:46PM +0200, Buchan Milne wrote:
However, now the "old" server is not getting data from clients, but the "new" one is, eg cpu, disk, files, memory, msgs, ports, procs are purple on the "old" server but green on the "new" sever (and collecting in the RRDs).
When proxy'ing the "client" messages, bbproxy only sends them to the first server listed in the --bbdisplay option.
The reason for this is that the client message gets back a response (the client configuration data). Since it only makes sense to send one of these responses back to the client, bbproxy only forwards the message to one server.
I'm not saying that is the right thing to do, because I really didn't think about it much. The "client" message support was added to bbproxy just like the other types of messages that get a response back ("query" and "config", for instance), but "client" is different because it actually sends data TO the Hobbit server, whereas the others just fetch some data FROM the Hobbit server.
So - I'll log this as a bug, and since there is also another problem with bbproxy already listed I'll see if it can be fixed.
Regards, Henrik
On Fri, Jul 14, 2006 at 05:46:11PM +0200, Henrik Stoerner wrote:
When proxy'ing the "client" messages, bbproxy only sends them to the first server listed in the --bbdisplay option.
This was in fact wrong - it sends it to the LAST server listed.
So - I'll log this as a bug, and since there is also another problem with bbproxy already listed I'll see if it can be fixed.
A patch is available now from the RC-patches page at http://www.hswn.dk/hobbitsw/betapatches/
Regards, Henrik
I've finished the first version of an extention to get data from netapp filers and get status alerts and data from some counters so that at the end I can be able to fully monitor a NetApp Filers in Hobbit. The script was based on the filers2bbstats.pl but use rsh command instead of SNMP data retrieval. This first version available on deadcat if someone need it get various data: cifs statistics, operations/seconds, iostats of tape/disk/net/fcp, snapshot volume list, qtree quotas, environmental check, snaplist and snap delta, snapmirror status, sysconfig, cpu usageand uptime, disk space and inode usage, users quotas.
For some of those kind of data I was able with ncv to create graphs (cifs statistics, ops/s and iostats). I want to get more data from the NetApp filers and graph them, an example are current snapshot dimension for each volume or vif statistics to know how the 4 aggregate 1gb interface are working and if they're splitting the net i/o in a right way.
The problem is that for each filer it's possible to have a different graph definition and a different dataset so that it's not so simple to implement something that will be user friendly.
One thing that can be useful could be the possibility to create nvc rrd file in a similar way to what the df modules already do (so 1 file for 1 dataset and the possibility to define a graph in the same way as the disk space graphs).
Another good way could be that of having the definitions in the status or data message in a way like:
vifstat data vif0:in:value vif0:out:value vif1:in:value vif1:out:value
and it will create something like vifstat.vif0.rrd with in and out value inside vifstat.vif1.rrd with in and out value inside
Thanks Francesco Duranti
participants (3)
-
bgmilne＠staff.telkomsa.net
-
fduranti＠q8.it
-
henrik＠hswn.dk