Xymon Mailing List Archive search

NCV handler problem

3 messages in this thread

list Matti Klock · Fri, 30 Sep 2005 11:19:34 -0400 ·
Hello,

I have an extension script called "send" that produces output
containing "seconds: <foo>", which I pass through the NCV handler to
create a graph. This works fine.

The problem comes when I add a link to the graph to the end of send's
output. It looks similar to <a href="http://$BBDISP/hobbit-cgi/stuff">;.
The graph appears on the test page, but data stops being collected, and
rrd-status.log fills up with errors:

2005-09-30 03:30:42 RRD error creating
/var/lib/hobbit/rrd/host/send.rrd: Duplicate DS name: http

I tried adding "http:NONE" to NCV_send in hobbitserver.cfg, and I tried
replacing the colon in "http://"; with %3A. Both actions changed the
errors to this:

2005-09-30 03:44:39 RRD error creating /var/lib/hobbit/rrd/host
/send.rrd: Duplicate DS name: 127001hobbitcgihobb

Is there a better way to get the NCV handler to ignore the colon in
http://? I'm running the snapshot from 21 Sep 2005 right now.

Thank you!

Matti
list Matti Klock · Fri, 30 Sep 2005 18:10:05 -0400 ·
quoted from Matti Klock
Matti Klock wrote:
Is there a better way to get the NCV handler to ignore the colon in
http://? I'm running the snapshot from 21 Sep 2005 right now.
Never mind, I figured it out. I'm having the same problem as Mario Andre
did on 23 Sep.

The NCV handler appears to have two bugs. First, it captures data of the
form <name>:<string>, as well as of the form <name>:<number>; second, it
reads beyond the end of the message sometimes, producing some really
bewildering errors in rrd-status.log.

This small patch fixed both problems, for what it's worth.

Thanks,
Matti
Attachments (1)
list Henrik Størner · Sat, 1 Oct 2005 08:23:52 +0200 ·
quoted from Matti Klock
On Fri, Sep 30, 2005 at 06:10:05PM -0400, Matti Klock wrote:
The NCV handler appears to have two bugs. First, it captures data of the
form <name>:<string>, as well as of the form <name>:<number>; second, it
reads beyond the end of the message sometimes, producing some really
bewildering errors in rrd-status.log.

This small patch fixed both problems, for what it's worth.
Looks correct, applied.  Thanks!


Henrik