Re: [collectd] Tail plugin and DSType gauge

2011-09-04 Thread Toni Ylenius
Hi Florian,

On Saturday 03 September 2011 05:43:48 Florian Forster wrote:
 Hi Toni,
 
 On Fri, Sep 02, 2011 at 04:20:17PM +0300, Toni Ylenius wrote:
  I found out that a value is reset to Nan in the function
  simple_submit_match (src/utils_tail_match.c) on each submit. I don't
  understand why is this, to me it makes more sense to retain the old
  value until the file is updated.
 
 I think there are two interesting cases:
 0) Several matches are found in the last x lines.
 1) No matches are found in those lines.
 
 How (0) is handled should be clear, I hope: In your case the minimum is
 computed and submitted.

That is true, but I can hardly imagine an application that writes some status 
information to a file many times within a short interval. However this may be 
useful, if plugins' intervals could be set individually.
 
 For (1) it is a bit tricky to define a good behavior: If you'd gather
 the average request size, for example, no requests means no average
 size, hence NaN (and not zero). In your case keeping the old value
 valid until another value has been matched might be preferable.
 
 I've whipped together a patch which implements this behavior. You can
 take a look at [0]. It's missing documentation and the new functionality
 should be ported to the curl and memcachec plugins -- preferably the
 configuration part would be de-duplicated in the process.

 Regards,
 —octo
 
 [0] https://collectd.org/gerrit/18

Thanks for your effort, the patch seems to work fine.

Toni

___
collectd mailing list
collectd@verplant.org
http://mailman.verplant.org/listinfo/collectd


Re: [collectd] Tail plugin and DSType gauge

2011-09-02 Thread Florian Forster
Hi Toni,

On Fri, Sep 02, 2011 at 04:20:17PM +0300, Toni Ylenius wrote:
 I found out that a value is reset to Nan in the function
 simple_submit_match (src/utils_tail_match.c) on each submit. I don't
 understand why is this, to me it makes more sense to retain the old
 value until the file is updated.

I think there are two interesting cases:
0) Several matches are found in the last x lines.
1) No matches are found in those lines.

How (0) is handled should be clear, I hope: In your case the minimum is
computed and submitted.

For (1) it is a bit tricky to define a good behavior: If you'd gather
the average request size, for example, no requests means no average
size, hence NaN (and not zero). In your case keeping the old value
valid until another value has been matched might be preferable.

I've whipped together a patch which implements this behavior. You can
take a look at [0]. It's missing documentation and the new functionality
should be ported to the curl and memcachec plugins -- preferably the
configuration part would be de-duplicated in the process.

Regards,
—octo

[0] https://collectd.org/gerrit/18
-- 
Florian octo Forster
Hacker in training
GnuPG: 0x0C705A15
http://octo.it/


signature.asc
Description: Digital signature
___
collectd mailing list
collectd@verplant.org
http://mailman.verplant.org/listinfo/collectd