[prev in list] [next in list] [prev in thread] [next in thread] 

List:       apache-httpd-users
Subject:    [users@httpd] Mod_proxy FailOnStatus based on error count
From:       Mandar Jog <mandarjog () gmail ! com>
Date:       2013-07-27 9:16:39
Message-ID: CA+GcQu9gCfOc89q89NrwsQ4PQAAQy4PFYOWsn7KKYLUPwYeTbA () mail ! gmail ! com
[Download RAW message or body]

Is there a way to mark a balancer worker in 'error state' based on count or
percentage of certain http errors.

If my cluster has 3 members, and one of them is resulting in a large number
of
500s then I would like to either mark the worker as 'bad' and remove it
from the cluster.

The number of errors has to be a percentage of total requests so that I can
distinguish between a spurious error 500 and a systemic problem affecting
the worker.

A better alternative to removing out of rotation is to significantly reduce
the weight of the member.

As I understand it, the current FailOnStatus is inadequate to handle this.

regards,
- Mandar

[Attachment #3 (text/html)]

<div dir="ltr"><div>Is there a way to mark a balancer worker in &#39;error state&#39; \
based on count or percentage of certain http errors.</div><div><br></div><div \
style>If my cluster has 3 members, and one of them is resulting in a large number \
of</div> <div style>500s then I would like to either mark the worker as &#39;bad&#39; \
and remove it from the cluster. </div><div style><br></div><div style>The number of \
errors has to be a percentage of total requests so that I can distinguish between a \
spurious error 500 and a systemic problem affecting the worker. </div> <div \
style><br></div><div style>A better alternative to removing out of rotation is to \
significantly reduce the weight of the member. <br></div><div style><br></div><div \
style>As I understand it, the current FailOnStatus is inadequate to handle this.<br> \
</div><div style><br></div><div style>regards,</div><div style>- Mandar</div></div>



[prev in list] [next in list] [prev in thread] [next in thread]