Xymon Mailing List Archive search

How to shorten delay before purple alerts?

6 messages in this thread

list Elizabeth Schwartz · Mon, 3 Oct 2011 09:35:24 -0400 ·
How can I change xymon to alert after a server has failed to report
for 15 minutes? We're finding that 30 minutes is too long a delay in
the production environment.

I see where you set purple delay for network tests and I see where you
can return it for individual tests, and I see where you can disable
purple. Is this a code change?

thanks Betsy
list Ralph Mitchell · Mon, 10 Oct 2011 09:42:47 -0400 ·
I finally had time to look around for this - it looks like the 30 minute
default is hard-wired into the code.

xymon-4.3.5/xymond/xymond.c, line 1210:

          int validity = 30;       /* validity is counted in minutes */

then at lines 1239-1241:

          if (strncmp(msg, "status+", 7) == 0) {
                validity = durationvalue(msg+7);
          }

So, an incoming message is given a validity of 30 minutes, unless set
otherwise by "status+X".

I just tested that by changing line 1210 to "validity=5", compiled,
installed, fired it up, sent a test status.  Five minutes later, I get a
purple dot for that test.

Ralph Mitchell


On Mon, Oct 3, 2011 at 9:35 AM, Elizabeth Schwartz <user-c61747246f66@xymon.invalid
quoted from Elizabeth Schwartz
wrote:
How can I change xymon to alert after a server has failed to report
for 15 minutes? We're finding that 30 minutes is too long a delay in
the production environment.

I see where you set purple delay for network tests and I see where you
can return it for individual tests, and I see where you can disable
purple. Is this a code change?

thanks Betsy

list Jeremy Laidman · Tue, 11 Oct 2011 16:16:50 +1100 ·
It would be really handy if this could be set by a variable within
xymonserver.cfg (eg XYMONDVALIDITY=30) and/or appending "--validity=N" to
the command line in tasks.cfg.  Using the environment variable might be
preferable in some cases, because it can be used by other things link
xymonnet or custom test scripts.
quoted from Ralph Mitchell

On Tue, Oct 11, 2011 at 12:42 AM, Ralph Mitchell <user-00a5e44c48c0@xymon.invalid>wrote:
I finally had time to look around for this - it looks like the 30 minute
default is hard-wired into the code.

xymon-4.3.5/xymond/xymond.c, line 1210:


          int validity = 30;       /* validity is counted in minutes */

then at lines 1239-1241:

          if (strncmp(msg, "status+", 7) == 0) {
                validity = durationvalue(msg+7);
          }

So, an incoming message is given a validity of 30 minutes, unless set
otherwise by "status+X".

I just tested that by changing line 1210 to "validity=5", compiled,
installed, fired it up, sent a test status.  Five minutes later, I get a
purple dot for that test.

Ralph Mitchell


On Mon, Oct 3, 2011 at 9:35 AM, Elizabeth Schwartz <
user-c61747246f66@xymon.invalid> wrote:
How can I change xymon to alert after a server has failed to report
for 15 minutes? We're finding that 30 minutes is too long a delay in
the production environment.

I see where you set purple delay for network tests and I see where you
can return it for individual tests, and I see where you can disable
purple. Is this a code change?

thanks Betsy

list Henrik Størner · Thu, 13 Oct 2011 14:23:13 +0200 ·
I've added a STATUSLIFETIME setting in xymonserver.cfg for the next
version.

Regards,
Henrik

On Tue, 11 Oct 2011 16:16:50 +1100, Jeremy Laidman
quoted from Jeremy Laidman
<user-71895fb2e44c@xymon.invalid> wrote:
It would be really handy if this could be set by a variable within
xymonserver.cfg (eg XYMONDVALIDITY=30) and/or appending "--validity=N"
to
the command line in tasks.cfg.  Using the environment variable might be
preferable in some cases, because it can be used by other things link
xymonnet or custom test scripts.

On Tue, Oct 11, 2011 at 12:42 AM, Ralph Mitchell
<user-00a5e44c48c0@xymon.invalid>wrote:
I finally had time to look around for this - it looks like the 30
minute
default is hard-wired into the code.

xymon-4.3.5/xymond/xymond.c, line 1210:


          int validity = 30;       /* validity is counted in minutes */

then at lines 1239-1241:

          if (strncmp(msg, "status+", 7) == 0) {
                validity = durationvalue(msg+7);
          }

So, an incoming message is given a validity of 30 minutes, unless set
otherwise by "status+X".

I just tested that by changing line 1210 to "validity=5", compiled,
installed, fired it up, sent a test status.  Five minutes later, I get
a
purple dot for that test.

Ralph Mitchell


On Mon, Oct 3, 2011 at 9:35 AM, Elizabeth Schwartz <
user-c61747246f66@xymon.invalid> wrote:
How can I change xymon to alert after a server has failed to report
for 15 minutes? We're finding that 30 minutes is too long a delay in
the production environment.

I see where you set purple delay for network tests and I see where you
can return it for individual tests, and I see where you can disable
purple. Is this a code change?

thanks Betsy

list Jeremy Laidman · Fri, 14 Oct 2011 12:11:55 +1100 ·
On Thu, Oct 13, 2011 at 11:23 PM,  <user-ce4a2c883f75@xymon.invalid> wrote:
I've added a STATUSLIFETIME setting in xymonserver.cfg for the next
version.
Great.  So I think that means I'll be able to do this in tasks.cfg, yes?:

  [xymonnet]
        CMD xymonnet --report --ping --checkresponse
--validity=${STATUSLIFETIME:-30}

Cheers
Jeremy
list Elizabeth Schwartz · Tue, 18 Oct 2011 09:45:58 -0400 ·
Cool thanks!

What's the timeline for the next release?

(I'm debating whether I should push a change control ticket to get
4.3.5 installed before our year-end freeze - if 4.3.6 is on the way I
might wait)