Xymon Network Tests - Behaviour?
Thank you for
Date: Wed, 24 Aug 2011 14:42:21 +0200 From: <henrik at hswn.dk> To: <xymon at xymon.com> Subject: Re: [Xymon] Xymon Network Tests - Behaviour? Message-ID: <04b91460cd235e5018e5179c06657153 at localhost> Content-Type: text/plain; charset=UTF-8
On Wed, 24 Aug 2011 13:52:55 +0200, Mario Valetti <mariov652 at gmail.com> wrote:
Could someone please explain the exact behaviour when a host becomes unreachable?
As I understand it, it works as follows...
Xymon pings each host only once,
if a reply is not received within a specified time (1sec), 3 ping requests are sent out,
from one of these three additional pings if a reply is received, the host is not indicated as being unavailable and things continue normally. If a reply is not received from these three additional requests, the host is marked as unavailable and four pings requests sent out every minute until a response is received.
Why do hosts show unavailable for 40s, 45s for example? Are the times above not fixed and rather work dynamically?
Not quite, but the normal test and the extra test are scheduled independently (they are two different tasks in tasks.cfg). The only restriction is that they will not run simultaneously. So when the first test detects that the host is down, the extra test may run anywhere from 0 to 59 seconds after that.
Second question is regarding hosts that are shutdown for maintenance, for example... Usually, these are set to 'until ok' and the host marked blue with no alerts. However Xymon still sends an arp request / ping to check when the host becomes available.
Is it possible to disable a host in a way that avoids Xymon checking for the host at all?
No. How would Xymon detect that the host becomes OK without trying to ping it ?
Regards, Henrik
Apologies, I'll try this again....
Thank you for all the responses to my orginal questions.
In my environment, I try to minimize unecassary traffic. My second question seems answered by simply adding the 'noping' in the bb-hosts file. This would work for my needs.
Regarding the first question, I'm currently using Xymon 4.3.0.0 beta 2. I'm assuming the tasks.cfg is used in the newer version(s) 4.3.3...? I'll also assume the behaviour you describe is the same with the [bbnet] and [bbretest] sections in the hobbitlaunch.cfg file in my version. Please let me know if I am wrong.
Thanks again, Mario
On 25/08/2011, Mario Valetti <mariov652 at gmail.com> wrote:
Thank you for
Date: Wed, 24 Aug 2011 14:42:21 +0200 From: <henrik at hswn.dk> To: <xymon at xymon.com> Subject: Re: [Xymon] Xymon Network Tests - Behaviour? Message-ID: <04b91460cd235e5018e5179c06657153 at localhost> Content-Type: text/plain; charset=UTF-8
On Wed, 24 Aug 2011 13:52:55 +0200, Mario Valetti <mariov652 at gmail.com> wrote:
Could someone please explain the exact behaviour when a host becomes unreachable?
As I understand it, it works as follows...
Xymon pings each host only once,
if a reply is not received within a specified time (1sec), 3 ping requests are sent out,
from one of these three additional pings if a reply is received, the host is not indicated as being unavailable and things continue normally. If a reply is not received from these three additional requests, the host is marked as unavailable and four pings requests sent out every minute until a response is received.
Why do hosts show unavailable for 40s, 45s for example? Are the times above not fixed and rather work dynamically?
Not quite, but the normal test and the extra test are scheduled independently (they are two different tasks in tasks.cfg). The only restriction is that they will not run simultaneously. So when the first test detects that the host is down, the extra test may run anywhere from 0 to 59 seconds after that.
Second question is regarding hosts that are shutdown for maintenance, for example... Usually, these are set to 'until ok' and the host marked blue with no alerts. However Xymon still sends an arp request / ping to check when the host becomes available.
Is it possible to disable a host in a way that avoids Xymon checking for the host at all?
No. How would Xymon detect that the host becomes OK without trying to ping it ?
Regards, Henrik
Last question...
Does this mean though, that the 'extra' test / bb-retest runs every minute but only sends out a ping/test if a host has been marked as unreachable?
Mario
On 25/08/2011, Mario Valetti <mariov652 at gmail.com> wrote:
Apologies, I'll try this again....
Thank you for all the responses to my orginal questions.
In my environment, I try to minimize unecassary traffic. My second question seems answered by simply adding the 'noping' in the bb-hosts file. This would work for my needs.
Regarding the first question, I'm currently using Xymon 4.3.0.0 beta 2. I'm assuming the tasks.cfg is used in the newer version(s) 4.3.3...? I'll also assume the behaviour you describe is the same with the [bbnet] and [bbretest] sections in the hobbitlaunch.cfg file in my version. Please let me know if I am wrong.
Thanks again, Mario
On 25/08/2011, Mario Valetti <mariov652 at gmail.com> wrote:
Thank you for
Date: Wed, 24 Aug 2011 14:42:21 +0200 From: <henrik at hswn.dk> To: <xymon at xymon.com> Subject: Re: [Xymon] Xymon Network Tests - Behaviour? Message-ID: <04b91460cd235e5018e5179c06657153 at localhost> Content-Type: text/plain; charset=UTF-8
On Wed, 24 Aug 2011 13:52:55 +0200, Mario Valetti <mariov652 at gmail.com> wrote:
Could someone please explain the exact behaviour when a host becomes unreachable?
As I understand it, it works as follows...
Xymon pings each host only once,
if a reply is not received within a specified time (1sec), 3 ping requests are sent out,
from one of these three additional pings if a reply is received, the host is not indicated as being unavailable and things continue normally. If a reply is not received from these three additional requests, the host is marked as unavailable and four pings requests sent out every minute until a response is received.
Why do hosts show unavailable for 40s, 45s for example? Are the times above not fixed and rather work dynamically?
Not quite, but the normal test and the extra test are scheduled independently (they are two different tasks in tasks.cfg). The only restriction is that they will not run simultaneously. So when the first test detects that the host is down, the extra test may run anywhere from 0 to 59 seconds after that.
Second question is regarding hosts that are shutdown for maintenance, for example... Usually, these are set to 'until ok' and the host marked blue with no alerts. However Xymon still sends an arp request / ping to check when the host becomes available.
Is it possible to disable a host in a way that avoids Xymon checking for the host at all?
No. How would Xymon detect that the host becomes OK without trying to ping it ?
Regards, Henrik
On 25-08-2011 16:25, Mario Valetti wrote:
Does this mean though, that the 'extra' test / bb-retest runs every minute but only sends out a ping/test if a host has been marked as unreachable?
Correct. The "normal" (once every 5 minutes) test drops a file in the Xymon "tmp" directory, with the servers that had some sort of failure for the network tests. The extra test picks up this file and retries the failed tests.
Regards, Henrik
participants (2)
-
henrik＠hswn.dk
-
mariov652＠gmail.com