We have situation where we are collecting a lot of rrd data for a specific test case (will be 16 data points). We'd like the ability to create two different graphs off of one rrd file (and one column on the hobbit page). We have a test called gt3_data that gets fed data from a script on a remote box via the bb command. This scripts sends up 8 different data values to be stored in the the gt3_data rrd. Currently we are graphing all the data points on one graph, but it is really ugly. We'd like to split it into two graphs, but we can't seem to find a way to tell hobbit to make two graphs based off of one column from the webpage. It doesn't make sense in our case to split the test into two different tests, but it does make sense to split the data into two graphs. Does anyone know if what this is possible?
Sean
On Tue, Jul 11, 2006 at 11:41:21AM -0400, Sean Hennessey wrote:
We have situation where we are collecting a lot of rrd data for a specific test case (will be 16 data points). We'd like the ability to create two different graphs off of one rrd file (and one column on the hobbit page). We have a test called gt3_data that gets fed data from a script on a remote box via the bb command. This scripts sends up 8 different data values to be stored in the the gt3_data rrd. Currently we are graphing all the data points on one graph, but it is really ugly. We'd like to split it into two graphs, but we can't seem to find a way to tell hobbit to make two graphs based off of one column from the webpage. It doesn't make sense in our case to split the test into two different tests, but it does make sense to split the data into two graphs. Does anyone know if what this is possible?
No, it's not possible with the current code.
It ought to be, I agree. But the whole handling of which graphs go where needs some major work; the current setup was adopted from Big Brother/LARRD and is not very flexible.
Regards, Henrik
Henrik Stoerner wrote:
On Tue, Jul 11, 2006 at 11:41:21AM -0400, Sean Hennessey wrote:
We have situation where we are collecting a lot of rrd data for a specific test case (will be 16 data points). We'd like the ability to create two different graphs off of one rrd file (and one column on the hobbit page). We have a test called gt3_data that gets fed data from a script on a remote box via the bb command. This scripts sends up 8 different data values to be stored in the the gt3_data rrd. Currently we are graphing all the data points on one graph, but it is really ugly. We'd like to split it into two graphs, but we can't seem to find a way to tell hobbit to make two graphs based off of one column from the webpage. It doesn't make sense in our case to split the test into two different tests, but it does make sense to split the data into two graphs. Does anyone know if what this is possible?
No, it's not possible with the current code.
It ought to be, I agree. But the whole handling of which graphs go where needs some major work; the current setup was adopted from Big Brother/LARRD and is not very flexible.
I thought you could do it via several calls to hobbitgraph? For example I have a host that has many disk partitions. If I don't want to see them all on a single graph, I can select just the first partition via the "first" and "count" options: https://www.myhost.com/hobbit-cgi/hobbitgraph.sh?host=myhost&service=disk&gr... This gives me a graph with only that data element. I can get a graph of only the 2nd via first=2&count=1, or elements 5-8, first=5&count=3.
Not the optimal way to do it, but it should work for creating a graph external to Hobbit, which won't have it built in as its own column the way you want, but you could always make calls into hobbitgraph from another html page.
-Charles
On Tue, Jul 11, 2006 at 02:08:16PM -0700, Charles Jones wrote:
Henrik Stoerner wrote:
On Tue, Jul 11, 2006 at 11:41:21AM -0400, Sean Hennessey wrote:
case to split the test into two different tests, but it does make sense to split the data into two graphs. Does anyone know if what this is possible?
No, it's not possible with the current code.
It ought to be, I agree. But the whole handling of which graphs go where needs some major work; the current setup was adopted from Big Brother/LARRD and is not very flexible.
I thought you could do it via several calls to hobbitgraph? For example I have a host that has many disk partitions. If I don't want to see them all on a single graph, I can select just the first partition via the "first" and "count" options: https://www.myhost.com/hobbit-cgi/hobbitgraph.sh?host=myhost&service=disk&gr... This gives me a graph with only that data element. I can get a graph of only the 2nd via first=2&count=1, or elements 5-8, first=5&count=3.
That only works if you store the data in multiple RRD files. The "count" parameter counts RRD files, not graph lines. (The "disk" rrd's only have one graph per file, so there's a one-to-one mapping in that case).
Regards, Henrik
Henrik Stoerner wrote:
On Tue, Jul 11, 2006 at 02:08:16PM -0700, Charles Jones wrote:
Henrik Stoerner wrote:
On Tue, Jul 11, 2006 at 11:41:21AM -0400, Sean Hennessey wrote:
case to split the test into two different tests, but it does make sense to split the data into two graphs. Does anyone know if what this is possible?
No, it's not possible with the current code.
It ought to be, I agree. But the whole handling of which graphs go where needs some major work; the current setup was adopted from Big Brother/LARRD and is not very flexible.
I thought you could do it via several calls to hobbitgraph? For example I have a host that has many disk partitions. If I don't want to see them all on a single graph, I can select just the first partition via the "first" and "count" options: https://www.myhost.com/hobbit-cgi/hobbitgraph.sh?host=myhost&service=disk&gr... This gives me a graph with only that data element. I can get a graph of only the 2nd via first=2&count=1, or elements 5-8, first=5&count=3.
That only works if you store the data in multiple RRD files. The "count" parameter counts RRD files, not graph lines. (The "disk" rrd's only have one graph per file, so there's a one-to-one mapping in that case).
Ah! Silly me I didn't realize that there is a seperate rrd for each partition. Ah well n00b is me :)
-Charles
participants (3)
-
henrik＠hswn.dk
-
jonescr＠cisco.com
-
sean.hennessey1＠verizonbusiness.com