Hobbit + MRTG
list Jeff Stuart
Ok, I've read the Hobbit + MRTG page. Now, question... I know I need to run mrtg on each machine that I'm monitoring BUT what I'm unsure of is how does Hobbit know about this data? IE do all the RRD files need to be copied over at some point to the display server? It's not very clear in the documentation as to if this scales to running MRTG on multiple boxes that you're already monitoring and if hobbit will actually display the graphs and if so, how specifically it will do it. -- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST
list Henrik Størner
▸
On Sun, May 21, 2006 at 05:01:24AM -0700, Jeff Stuart wrote:
Ok, I've read the Hobbit + MRTG page. Now, question... I know I need to run mrtg on each machine that I'm monitoring BUT what I'm unsure of is how does Hobbit know about this data? IE do all the RRD files need to be copied over at some point to the display server? It's not very clear in the documentation as to if this scales to running MRTG on multiple boxes that you're already monitoring and if hobbit will actually display the graphs and if so, how specifically it will do it.
The only thing Hobbit does about MRTG is to use the RRD files generated by MRTG and show them on the "trends" page for each host. So yes, those files will need to be copied over to the Hobbit server, specifically to the ~hobbit/data/rrd/HOSTNAME/ directory. Regards, Henrik
list Jeff Stuart
Ok, that begs the question then. Is anyone here using Hobbit + MRTG to monitor bandwidth of multiple servers? If so, how are you doing it? IIRC, MRTG uses snmp to grab the data from the machine. Is it possible to have mrtg on one machine poll many boxes every 5 minutes? Or is it better to copy the files over to the display server every so often? Finally, what about a client ext script that sent the current bandwidth and then hobbit stores it as a CSV graph or just as a standard external script + rrd data file?
▸
On Sunday 21 May 2006 5:19 am, Henrik Stoerner wrote:On Sun, May 21, 2006 at 05:01:24AM -0700, Jeff Stuart wrote:Ok, I've read the Hobbit + MRTG page. Now, question... I know I need to run mrtg on each machine that I'm monitoring BUT what I'm unsure of is how does Hobbit know about this data? IE do all the RRD files need to be copied over at some point to the display server? It's not very clear in the documentation as to if this scales to running MRTG on multiple boxes that you're already monitoring and if hobbit will actually display the graphs and if so, how specifically it will do it.The only thing Hobbit does about MRTG is to use the RRD files generated by MRTG and show them on the "trends" page for each host. So yes, those files will need to be copied over to the Hobbit server, specifically to the ~hobbit/data/rrd/HOSTNAME/ directory.
-- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST
list Sue Bauer-Lee
I'd still really like to see a successful configuration that monitors Cisco routers and switches. I've yet to be able to display the graphs for the few with successful data collection; most of them fail and we're fairly certain it has nothing to do with ACLs. I was able to collect data for, which are few, and mo
▸
On Sun, May 21, 2006 at 05:33:27AM -0700, Jeff Stuart wrote:Ok, that begs the question then. Is anyone here using Hobbit + MRTG to monitor bandwidth of multiple servers? If so, how are you doing it? IIRC, MRTG uses snmp to grab the data from the machine. Is it possible to have mrtg on one machine poll many boxes every 5 minutes? Or is it better to copy the files over to the display server every so often? Finally, what about a client ext script that sent the current bandwidth and then hobbit stores it as a CSV graph or just as a standard external script + rrd data file? On Sunday 21 May 2006 5:19 am, Henrik Stoerner wrote:On Sun, May 21, 2006 at 05:01:24AM -0700, Jeff Stuart wrote:Ok, I've read the Hobbit + MRTG page. Now, question... I know I need to run mrtg on each machine that I'm monitoring BUT what I'm unsure of is how does Hobbit know about this data? IE do all the RRD files need to be copied over at some point to the display server? It's not very clear in the documentation as to if this scales to running MRTG on multiple boxes that you're already monitoring and if hobbit will actually display the graphs and if so, how specifically it will do it.The only thing Hobbit does about MRTG is to use the RRD files generated by MRTG and show them on the "trends" page for each host. So yes, those files will need to be copied over to the Hobbit server, specifically to the ~hobbit/data/rrd/HOSTNAME/ directory.-- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST
list Jeff Stuart
Hmm ok thx... looks like this'll be my next big project. ;) (Last one was to integrate hobbit with our server provisioning system/database. IE adds/deletes add and delete hosts in hobbit via a couple of perl scripts.)
▸
On Sunday 21 May 2006 5:41 am, Sue Bauer-Lee wrote:I'd still really like to see a successful configuration that monitors Cisco routers and switches. I've yet to be able to display the graphs for the few with successful data collection; most of them fail and we're fairly certain it has nothing to do with ACLs. I was able to collect data for, which are few, and mo
-- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST
list Robert Taylor
I use MRTG and Hobbit to monitor routers and switches. My config basically is a follows: 1. bbmrtg.sh runs on a host that collects all of our bandwidth information. 2. bbmrtg.sh sends the information to Hobbit. 3. Hobbit displays it as a mrtg column similar to the way that it used to work in Big Brother. I have reconfigured bbmrtg.sh to work with the 14all.cgi application instead of the original configuration as 14all.cgi allowed me to display performance in a per-minute graph as opposed to the 5 minute graph that was originally displayed. The mrtg column shows a 6 hour bandwidth view and still alerts based on the pre-defined thresholds. If there is a particular host that I want to represent the bandwidth for then I just add that port to the bbmrtg.sh configuration file. If I could get Hobbit to display graphs on finer detail for bandwidth then I would likely move things off of the mrtg host that I am using now, but I just haven't given it a good look. The other main reason that I like bbmrtg.sh is that I don't have to mess with any of my mrtg configuration files. From what I have seen in other configurations I have to go into the mrtg files and edit them individually before Hobbit would recognize them. I could be wrong on this, but that is what I recall. I have left things this way as opposed to moving them into the trends section of Hobbit as I prefer to have 1 minute averages versus the 5 minute ones as it is a little easier to spot problem situations when there is greater granularity in the graph. Robert Taylor | Senior Manager of Information Systems | Hendrick Automotive Group
▸
-----Original Message-----
From: Jeff Stuart [mailto:user-968ca3722b80@xymon.invalid]
Sent: Sunday, May 21, 2006 9:15 AM
To: user-ae9b8668bcde@xymon.invalid
Subject: Re: [hobbit] Hobbit + MRTG
Hmm ok thx... looks like this'll be my next big project. ;) (Last one
was to
integrate hobbit with our server provisioning system/database. IE
adds/deletes add and delete hosts in hobbit via a couple of perl
scripts.)
On Sunday 21 May 2006 5:41 am, Sue Bauer-Lee wrote:I'd still really like to see a successful configuration that monitors Cisco routers and switches. I've yet to be able to display the graphs for the few with successful data collection; most of them fail and we're fairly certain it has nothing to do with ACLs. I was able to collect data for, which are few, and mo
-- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST
list Jeff Stuart
Hmm.. ok.. here's my layout. We've got 4 nocs setup across the US. We've got bunches of servers in each noc. We've also got a few resellers who have boxes in one or more nocs. Here's what I want to do: 1) Setup a NET testing server in each noc. That would probably also run the bbmrtg.sh script and send both the net tests AND the bandwidth data to the display server. 2) Have a separate URL for each reseller so that they can see the status of their own servers. 3) POSSIBLY also do external testing of all the servers. (IE for each NOC, it not only does the internal net tests but will do say pings of all the servers in the other nocs.) This data then gets merged so that I have an overall view of how things are going across the US. (IE we have nocs in California, Texas, Virginia, and New York so kinda gives us a good feel how the net in general is doing. hehe) 4) Also test the main router at each NOC and if IT'S down, then of course don't do prop bad statuses of all the servers behind the router. How to do these things: 1) I ASSUME this means I just setup a hobbit server on each of the net testers but I do NOT enable anything but the hobbitd, bbnet, bbreset, hobbitclient, clientdata, and bbmrtg programs in hobbitlaunch.cfg. This would also use the NET:location keyword for each host + defining BBLOCATION on each BBNET server. And in the hobbitserver.cfg make sure that my bbdisplay IP (plus other data) all points to my display server. 2) This is done via pagesets I assume? (If so, I understand how this works already. :)) 3) This one... I have no clue if it's possible or even IF advisable... 4) I assume this will require that I use the depends and route keywords for each host set appropriately. Does all this make sense? Am I understanding how this thing works? :) Is any of this doable/feasible? P.S. If anyone is doing anything like any of the above, I'd LOVE to see sample configs if possible just to get an idea on how things need to be setup properly.
▸
On Sunday 21 May 2006 10:13 am, Taylor, Robert wrote:I use MRTG and Hobbit to monitor routers and switches. My config basically is a follows: 1. bbmrtg.sh runs on a host that collects all of our bandwidth information. 2. bbmrtg.sh sends the information to Hobbit. 3. Hobbit displays it as a mrtg column similar to the way that it used to work in Big Brother. I have reconfigured bbmrtg.sh to work with the 14all.cgi application instead of the original configuration as 14all.cgi allowed me to display performance in a per-minute graph as opposed to the 5 minute graph that was originally displayed. The mrtg column shows a 6 hour bandwidth view and still alerts based on the pre-defined thresholds. If there is a particular host that I want to represent the bandwidth for then I just add that port to the bbmrtg.sh configuration file. If I could get Hobbit to display graphs on finer detail for bandwidth then I would likely move things off of the mrtg host that I am using now, but I just haven't given it a good look. The other main reason that I like bbmrtg.sh is that I don't have to mess with any of my mrtg configuration files. From what I have seen in other configurations I have to go into the mrtg files and edit them individually before Hobbit would recognize them. I could be wrong on this, but that is what I recall. I have left things this way as opposed to moving them into the trends section of Hobbit as I prefer to have 1 minute averages versus the 5 minute ones as it is a little easier to spot problem situations when there is greater granularity in the graph. Robert Taylor | Senior Manager of Information Systems | Hendrick Automotive Group
-- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST
list Jeff Stuart
Forgot to add that for the main page I want to see ALL servers include all of the reseller boxes. I don't remember if pagesets do what I want or not still. (I THINK they do but I don't remember.)
▸
On Monday 22 May 2006 1:38 am, Jeff Stuart wrote:2) This is done via pagesets I assume? (If so, I understand how this works already. :))
-- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST
list Paul Williamson
user-968ca3722b80@xymon.invalid 05/21/06 8:01 AM >>>
▸
Ok, I've read the Hobbit + MRTG page. Now, question... I know I need
to run mrtg on each machine that I'm monitoring BUT what I'm unsure of is how does Hobbit know about this data? IE do all the RRD files need to be copied over at some point to the display server?
WHAT? MRTG only needs to be running on the collection server. SNMP needs to be running on the remote machine. You need to read some MRTG docs to figure out what MRTG can do, and there is no reason to run MRTG on every machine you monitor. I'd get shot if I asked to run MRTG on all 10,000 or so servers I'm monitoring... Paul
list Jeff Stuart
Ok, that is a bit more in line with what I thought how MRTG worked. Figure if I run one collection server per noc, i'll be good to go.
▸
On Monday 22 May 2006 5:43 am, PAUL WILLIAMSON wrote:user-968ca3722b80@xymon.invalid 05/21/06 8:01 AM >>>Ok, I've read the Hobbit + MRTG page. Now, question... I know I need to run mrtg on each machine that I'm monitoring BUT what I'm unsure of is how does Hobbit know about this data? IE do all the RRD filesneedto be copied over at some point to the display server?WHAT? MRTG only needs to be running on the collection server. SNMP needs to be running on the remote machine. You need to read some MRTG docs to figure out what MRTG can do, and there is no reason to run MRTG on every machine you monitor. I'd get shot if I asked to run MRTG on all 10,000 or so servers I'm monitoring... Paul
-- Jeff Stuart Network Admin MyInternetServices.com 1-800-300-HOST