Graphs For MongoDB

[GoogleCodeExporter]

Hi All,

I am currently working with the MongoDB Document Store 
(http://www.mongodb.org/display/DOCS/Home) and needed Cacti performance 
graphs. I used your _excellent_ framework to create graphs, data sources 
and unit tests for Connected Clients, Index Ops, Memory, Slave Lag and 
pretty much everything else exposed through the systemStatus command in 
MongoDB.

I'm going to release these today for public consumption and figured I'd 
post them here as well in case you wanted to include them. I'd be happy to 
maintain them if wanted.

I cannot thank you all enough for this framework, I've created graphs by 
hand before and god damn is this nicer.

Anyway, the diff from svn TRUNK as of a few days ago is attached. Let me 
know if your interested in incorporating these.

-
Narayan Newton
Tag1 Consulting

Original issue reported on code.google.com by narayann...@gmail.com on 31 Mar 2010 at 8:41

Attachments:

• mongo_graphs.diff

[GoogleCodeExporter]

That's awesome.  I just gave you SVN commit rights.  I think it's ideal to 
maintain
these through this project where there's already a framework to support it all 
--
issue tracker, releases, etc.  So, please go ahead and commit, and I'll then 
I'll
also do a code review and give feedback.  I'm working on a couple things right 
now
and soon it'll be time to roll another release.

Original comment by baron.schwartz on 1 Apr 2010 at 12:53

• Changed state: Accepted
• Added labels: Type-Enhancement
• Removed labels: Type-Defect

[GoogleCodeExporter]

Great, thanks. I'll try to get these integrated today.

-N

Original comment by narayann...@gmail.com on 1 Apr 2010 at 4:52

[GoogleCodeExporter]

Removing unused labels.

Original comment by baron.schwartz on 16 Apr 2010 at 2:14

• Removed labels: Priority-Medium

[GoogleCodeExporter]

My apologies it took so long to get to this, I've finally added in these graphs 
and 
forward ported them to 1.1.7, to fit in with the jmx graphs. Todo is I need to 
add a 
slow queries graph and I believe I should change a few of the colors. Revision 
is 502.

Looking forward to your comments.

Original comment by narayann...@gmail.com on 21 Apr 2010 at 10:38

[GoogleCodeExporter]

Awesome.  Can you add a wiki page (just note at the top that the graphs are new 
and
will be released in version 1.1.8), and attach sample graphs to it?

Original comment by baron.schwartz on 21 Apr 2010 at 11:06

[GoogleCodeExporter]

Baron-

Absolutely. Sorry for the delay, we have been tracking down a rather critical 
MongoDB 
debug in the 1.4 release. 

-N

Original comment by narayann...@gmail.com on 5 May 2010 at 11:35

[GoogleCodeExporter]

Great!  Thanks for adding that.  I am looking at the sample file and I see this:

   "globalLock" : {
      "totalTime" : 288777173126,
      "lockTime" : 1366671934,
      "ratio" : 0.0047326176068760465
   },

Do you think that's worth graphing?

Original comment by baron.schwartz on 6 May 2010 at 12:30

[GoogleCodeExporter]

The MongoDB docs are less than revealing about what those numbers really mean.  
I've
taken this to the mongo mailing list to see if I can find out.

Original comment by baron.schwartz on 6 May 2010 at 1:36

[GoogleCodeExporter]

This is actually on my list of things to graph, but a graph of slow queries was 
higher 
on the priority. This maybe easier though. I have a question into Eliot as well 
about 
this.

-N

Original comment by narayann...@gmail.com on 6 May 2010 at 6:01

[GoogleCodeExporter]

And got an answer...isn't having a support contract fun. These numbers are the 
total 
query run time for any query and the total time any query has waited for the 
global 
write lock. So this is actually very useful to graph. I'll try to add this 
tonight.

-N

Original comment by narayann...@gmail.com on 6 May 2010 at 6:03

[GoogleCodeExporter]

And got an answer...isn't having a support contract fun. These numbers are the 
total 
query run time for any query and the total time any query has waited for the 
global 
write lock. So this is actually very useful to graph. I'll try to add this 
tonight.

-N

Original comment by narayann...@gmail.com on 6 May 2010 at 6:03

[GoogleCodeExporter]

Was also pointed to this: http://github.com/erh/mongo-munin

Two graphs there I want to rewrite in this framework.

-N

Original comment by narayann...@gmail.com on 6 May 2010 at 6:06

[GoogleCodeExporter]

The answer I got about the lock time numbers was different, and actually made 
more
sense to me.  From
http://groups.google.com/group/mongodb-user/browse_thread/thread/a053c3168b8bb24
9

totalTime is uptime
lockTime is the amount of time the global lock has been held 

In checking through that, it looks like it's correct.  More details are there.

In my experience, graphing just lockTime is probably best; totalTime is just 
going to
be a flat line!

Original comment by baron.schwartz on 7 May 2010 at 12:59

[GoogleCodeExporter]

Sorry, I explained that poorly. We got the same answer, mine is just different 
because 
I asked a specific question about the reader/writer lock and explained it 
extremely 
poorly. I agree lockTime is what should be graphed.

-N

Original comment by narayann...@gmail.com on 7 May 2010 at 3:14

[GoogleCodeExporter]

narayannewton, I am thinking about making a new release soon.  Do you have any 
futher enhancements for these templates, or should we release them now and 
update them later if there is more to add?

Original comment by baron.schwartz on 15 Jan 2011 at 5:44

[GoogleCodeExporter]

Hi Baron,

These have been working nicely so far for me. I should finish the 
documentation, but as for the graphs themselves I don't have immediate plans.

-N

Original comment by narayann...@gmail.com on 18 Jan 2011 at 11:53

[GoogleCodeExporter]

Great.  I'm going to release them and blog about them so we can get more 
attention.  Maybe even the MongoDB newsletter will mention them; they seem to 
be really good about mentioning community stuff that goes on.

Original comment by baron.schwartz on 19 Jan 2011 at 3:26

• Changed state: Fixed