[Reconnoiter-users] problem rolling up metrics

Theo Schlossnagle jesus at omniti.com
Wed Jun 2 11:07:28 EDT 2010


'C' log messages are emitted at startup and any time a check configuration "changes" which includes adds.  so, if you restart noitd, a 'C' message should be emitted again and inserted into check_currently.

On Jun 2, 2010, at 1:18 AM, Gareth Stokes wrote:

> ahh riight, 
> 
> well my stratcon is not seeing any 'C' log messages. my output consists of the one collectd check and thats it.
> 
> Pushing durable/storage checkpoint [noit-test]: [1/10342]
> Pushing transient/iep batch async [noit-test]: [1/10343]
> Pushing transient/iep checkpoint [noit-test]: [1/10343]
> AMQOutput -> dispatch
> {
>   "r": {
>     "availability": "A",
>     "duration": 29.0,
>     "length": 8,
>     "noit": "127.0.0.1",
>     "prefix": "S",
>     "state": "G",
>     "status": "dur=29,run=1,stats=145,ntfy=0",
>     "time": 1275455555034,
>     "uuid": "12c6761e-d16f-4f33-9451-f2f275d7cca1"
>   }
> }
> 
> 
> 
> On 1 June 2010 23:35, Theo Schlossnagle <jesus at omniti.com> wrote:
> The check_currently should be populated directly by stratcon every time it sees the 'C' log line.
> 
> The metric_name_summary requires the check to be in check_currently and will populate via triggers on noit.metric_numeric_archive and noit.metric_text_archive.
> 
> On Jun 1, 2010, at 2:33 AM, Gareth Stokes wrote:
> 
> > ok, i think im missing some triggers in my postgres database.
> >
> > when i insert into check_status_archive_20100601 or similar, i have the following trigger getting fired ...
> > check_status_archive_log_changes
> >
> > when i look into that, all i see happening is a check against the check_status_changelog table with a conditional insert that doesn't get fired due to the fact that the state or availability hasn't changed.
> >
> > what should i be looking for to get the data i need into the check_currently and metric_name_summary tables?
> >
> > On 31 May 2010 16:16, Theo E. Schlossnagle <jesus at omniti.com> wrote:
> > This is a really annoying bug that is trigger if you get yournoits up and running before your database is working right. It should happen only once and a restart of the noits should soave the problem.
> >
> > --
> > Theo Schlossnagle (mobile)
> > http://omniti.com/is/theo-schlossnagle
> >
> > On May 31, 2010, at 12:46 AM, Gareth Stokes <gareth at betechnology.com.au> wrote:
> >
> > > Hey,
> > >
> > > So i've got reconoiter configured to accept metrics from collectd and the noits / stratcon are working quite well. If i tail the postgres logs i can see the insert into the table : check_status_archive_[date]
> > >
> > > also i can see the following query being called from the cron task
> > >
> > > select stratcon.rollup_metric_numeric(rollup) from metric_numeric_rollup_config order by seconds asc;
> > >
> > > the problem is that the web page is displaying no stats and it looks like its needing data inside the check_currently table, when i look inside there its empty?
> > >
> > > what am i missing here?
> > >
> > > cheers,
> > > gareth stokes
> > >
> > > _______________________________________________
> > > Reconnoiter-users mailing list
> > > Reconnoiter-users at lists.omniti.com
> > > http://lists.omniti.com/mailman/listinfo/reconnoiter-users
> >
> 
> --
> Theo Schlossnagle
> http://omniti.com/is/theo-schlossnagle
> 
> 
> 
> 
> 
> 

--
Theo Schlossnagle
http://omniti.com/is/theo-schlossnagle








More information about the Reconnoiter-users mailing list