[rrd-developers] [PATCH] rrdcached "SLURP" for extreme update rates

kevin brintnall kbrint at rufus.net
Fri Oct 3 21:01:58 CEST 2008


> >     ->	MULTIUPDATE
> >     <- 	0 Go ahead.  Terminate with '.' on its own line (is this necesary?)
> >     ->	x.rrd 1:2
> >     -> 	y.rrd 2:3
> >     -> 	.
> >     <-	2 results
> >     <- 	x.rrd ??
> >     <-	y.rrd ??
> >
> > A few questions:
> >
> >  ? Should we do without the initial "0 Go ahead"?
> 
> Maybe get inspiration from SMTP ?

That's what I did :)

> I think a termination command makes sense from a 'language' design
> point of view, since otherwhise the rrdtool update lines become
> commands in their own right.

I think we only need a terminator on multi-line commands from the client.

> >  ? how do we match up updates/responses?
> >    - if using a separate key, how do we separate from the filename?
> >    - what if the user doesn't care about response? (empty key?)
> 
> I would only send back complaints, input linenumber followed by the
> error. The client can then make sense of it since it can keep track
> of the stuff it sent ...

Good idea...  simple.

> this brings up the question of how to react on an error. Will the
> deamon ignore further updates to the same rrd but continue
> processing updates to other rrds, or will it abort completely ?

On any permanent error (i.e. file doesn't exist) it won't matter.

Statement-specific errors (i.e. bad command) will just be ignored.

RRD-specific errors (i.e. backward timestamp)...  we don't do any
validation today, so we'll have to figure that out first.

-- 
 kevin brintnall =~ /kbrint at rufus.net/



More information about the rrd-developers mailing list