[collectd] table plugin and match

[Bill Schwanitz]

Question on the Table plugin and use of Match

I am trying to collect per numa-node meminfo stats. For some reason, every server we have has the first line of the meminfo file ( only in a numa node ) has a blank line followed by the typical meminfo data.

 <Table "/sys/devices/system/node/node0/meminfo">
   Instance "numa_meminfo"
   Separator " "
   <Result>
     Type gauge
     InstancePrefix "node0"
     InstancesFrom 2
     ValuesFrom 3
   </Result>
 </Table>

Oct  6 19:40:04 foo collectd[3260]: table plugin: Not enough columns in line (expected at least 4, got 1).
Oct  6 19:40:04 foo collectd[3260]: table plugin: Table /sys/devices/system/node/node1/meminfo: Failed to parse line: 

I have tried putting a Match block in both the top level Table ( below Separator ) and within the Result section - neither place works and I get an error

Oct  6 19:29:55 foo collectd[15596]: table plugin: Ignoring unknown config key "Match" in <Table /sys/devices/system/node/node0/meminfo>.

   <Match> 
     Regex "^Node"
   </Match>

Here is the firt 5 lines of the meminfo file

-- begin
me at foo:~$ head -5 /sys/devices/system/node/node1/meminfo

Node 1 MemTotal:       12582912 kB
Node 1 MemFree:         2888544 kB
Node 1 MemUsed:         9694368 kB
Node 1 Active:          4361740 kB
-- end

Is there a way to either quiet the table plugin or get a match looking for lines starting with Node? I don't mind switching to an exec script but I'd rather not.

Bill Schwanitz

--

"Mindlessness and moral idiocy are not characteristically human attributes; they are symptoms of herd-poisoning."
 -- Aldous Huxley, Brave New World Revisited ( pg. 44 )
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.verplant.org/pipermail/collectd/attachments/20121007/bc3edf16/attachment.html>