Xymon Mailing List Archive search

Bogus hobbitfetch reds and purples

3 messages in this thread

list Hobbit User · Mon, 4 Jun 2007 08:38:46 -0400 (EDT) ·
I have hobbitfetch working nicely for data fetching, but it continually
goes red, then purple, with a detail of "-Program crashed" and "Fatal
signal caught!".  When this  happens, it's still fetching the client data
without a hitch as far as I can see.  When I came in this morning, I had
the purple, with "Status unchanged in 0 hours 22 minutes", since I started
writing this, it's gone to red with "Status unchanged in 0 hours 3
minutes", all the while happily updating the client data.

hobbitfetch.log has 0 length and was last touched about four hours ago. 
hobbitlaunch.log says "hobbitlaunch.log:2007-06-04 08:34:22 Task
hobbitfetch terminated by signal 6".  The client data display available
through the hobbitfetch status page is stamped 8:32:35 and clearly shows
the hobbitfetch process running.

Clues?
list Daniel J McDonald · Mon, 04 Jun 2007 07:49:53 -0500 ·
quoted from Hobbit User
On Mon, 2007-06-04 at 08:38 -0400, Hobbit User wrote:
I have hobbitfetch working nicely for data fetching, but it continually
goes red, then purple, with a detail of "-Program crashed" and "Fatal
signal caught!".  When this  happens, it's still fetching the client data
without a hitch as far as I can see.  When I came in this morning, I had
the purple, with "Status unchanged in 0 hours 22 minutes", since I started
writing this, it's gone to red with "Status unchanged in 0 hours 3
minutes", all the while happily updating the client data.

hobbitfetch.log has 0 length and was last touched about four hours ago. 
hobbitlaunch.log says "hobbitlaunch.log:2007-06-04 08:34:22 Task
hobbitfetch terminated by signal 6".  The client data display available
through the hobbitfetch status page is stamped 8:32:35 and clearly shows
the hobbitfetch process running.

Clues?
Be glad it doesn't just sit and consume 100% cpu, like it does a couple
of times a day on my box.  At that point, hobbitfetch turns yellow and
all of the tests being fetched turn purple.

I think it is horribly broken, but I have no workaround, so I'm up for
killing the hobbitfetch process two or three times a day...


-- 
Daniel J McDonald, CCIE # 2495, CISSP # 78281, CNX
Austin Energy
http://www.austinenergy.com
list Hobbit User · Mon, 4 Jun 2007 12:15:59 -0400 (EDT) ·
quoted from Daniel J McDonald
On Mon, June 4, 2007 08:49, Daniel J McDonald wrote:
Be glad it doesn't just sit and consume 100% cpu, like it does a couple
of times a day on my box.  At that point, hobbitfetch turns yellow and
all of the tests being fetched turn purple.

I think it is horribly broken, but I have no workaround, so I'm up for
killing the hobbitfetch process two or three times a day...
Hoo, boy.  I'm getting queasier about this all the time.  Since the
hobbitfetch alert results were meaningless, I put in a
NOCOLUMNS:hobbitfetch and did a bb "drop host hobbitfetch".  The column
went away, and then CAME BACK SPONTANEOUSLY!  AiggggHHHH!  Oh, wait a
minute, hobbitfetch is not in the list of supported columns for NOCOLUMNS.
 I'm not sure I can stomach doing the column drop every minute in a cron
job, so lessee, maybe: group-except hobbitfetch Hobbit Display
Server...yeah, yeah, that's it....