Hi all
Well, it's been 48 hours now, without an incident.
Seems that not all fpings are created equal.
I did install the OpenCSW fping when I installed Xymon. The the install
script defaults to using the last instance of fping it finds, and looks for
the OpenCSW fping last.
From build/fping.sh
echo "Checking for fping ..."
for DIR in / /usr /usr/local /opt /usr/pkg /opt/csw
do
if test "$DIR" = "/"; then DIR=""; fi
if test -x $DIR/bin/fping
then
FPING=$DIR/bin/fping
elif test -x $DIR/sbin/fping
then
FPING=$DIR/sbin/fping
fi
done
I would have expected this to default to the OpenCSW version of fping.
So what went wrong?
My recollection of the install is poor, but It is possible that I ran the
configure script as Xymon user, and had only set the suid bit on
/usr/local/sbin/fping, meaning that
test -x $DIR/bin/fping
would have failed for the OpenCSW version, defaulting to the fping in
/usr/local.
Of course, the script would have reported no errors, because it would have
met all requirements. An executable fping.
*Solution *: Make sure that Xymon is using the OpenCSW fping, as defined by
the FPING= variable in ~/server/etc/xymonserver.cfg
*Prevention 1 *: When running configure at install time, use the --fping
option to explicitly define which fping to use.
*Prevention 2* : Make sure you set the suid bit on ALL instances of fping.
Regards
Vernon
On 15 January 2013 08:27, Vernon Everett <user-b3f8dacb72c8@xymon.invalid> wrote:
Hi all
"TIME SPENT" is showing the odd spike now and then, but the highest I have
seen is 9.8 seconds.
And these peaks do not coincide with the fping issues.
I did a little more research, and discovered 2 instances of fping.
# /opt/csw/sbin/fping -v
/opt/csw/sbin/fping: Version 2.4b2_to $Date: 2002/01/16 00:33:42 $
/opt/csw/sbin/fping: comments to user-eaf9c6286157@xymon.invalid
# /usr/local/sbin/fping -v
/usr/local/sbin/fping: Version 3.0
/usr/local/sbin/fping: comments to user-359c54915310@xymon.invalid
/opt/csw/sbin/fping is the one I installed recently using OpenCSW's
pkgutil, but the Xymon config was pointing to /usr/local/sbin/fping
I have switched the FPING= definition in ~/server/etc/xymonserver.cfg to
point to the OpenCSW version.
Let's see if this improves matters.
Regards
Vernon
On 11 January 2013 08:16, Jeremy Laidman <user-71895fb2e44c@xymon.invalid> wrote:
I think the key here is the "select". To me, this suggests that the
select() system call (which is used to read data on STDIN from xymonnet) is
trying to get data from a closed socket, or a similar problem. I wonder if
fping, which is run from xymonnet, is sometimes taking so long to run, and
eventually xymonlaunch is killing xymonnet.
By default, xymonlaunch (in tasks.cfg) kills xymonnet after 5 minutes,
but xymonnetagain is killed off after only 1 minute. So if you have a lot
of hosts that are uncontactable, or have long ping times, then this might
add up to a minute.
To confirm this, check out the "xymonnet" dot on your Xymon server, and
look at the "TIME SPENT" table for ping test duration.
J
On 10 January 2013 17:12, Vernon Everett <user-b3f8dacb72c8@xymon.invalid>wrote:
Hi all
Anybody seen something like this?
Execution of '/usr/local/sbin/fping -Ae' failed with error-code 4
/usr/local/sbin/fping: select : Invalid argument
Error I see when xymonnet page goes red.
At the same time, all conn tests go white.
Happens briefly every few hours. No obvious pattern, but at least once a
day, sometimes twice.
This is a new installation of Xymon 4.3.10, on a relatively old Solaris
10 Sparc server.
Regards
Vernon
--
"Accept the challenges so that you can feel the exhilaration of victory"
- General George Patton
--
"Accept the challenges so that you can feel the exhilaration of victory"
- General George Patton
--
"Accept the challenges so that you can feel the exhilaration of victory"
- General George Patton