[rrd-users] RRD Cached Daemon not flushing properly (queue is growing fast)

[Jose Oscar Olivares Ocampo/Mexico/IBM]

     Hello Everyone, we have some problems and we are running out of 
suggestions, so anything you can add will be highly appreciated.

We have a process that runs every hour, it collects a lot of information 
and send it to the cached daemon for update. An update is done as soon as 
something has been collected. But we have several rrd files being handled 
by the daemon, I mean, the before mentioned update is not done to a single 
file, but depending on the collected information, it updates one specific 
rrd file.
There are sometimes when queue grows a lot (more than 200, 000 entries). 
Since this process runs every hour, it not always end flushing the data in 
an hour.

I believe (but I am not totally certain about it) that the problem occurs 
when one of the process takes longer than an hour to complete, such that 
at certain point, there are 2 processes sending information to the daemon.
Is there any special consideration for this case? Is there any special way 
to treat the daemon under this circumstance?

And that brings up another question. How does the daemon works when 
flushing his data? Sometimes it appears that it first flush the new 
information (the one that has recently arrived), instead of the old 
information (information that had arrived a lot of minutes ago - sometimes 
even an hour ago). Is this a normal behavior? Or are we missing something 
here?

Thanks in Advanced.

Best Regards!

José Oscar Olivares Ocampo
Infrastructure Tools
Tie Line: 877-3115
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.oetiker.ch/pipermail/rrd-users/attachments/20100601/15641784/attachment-0001.htm