LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: monitoring tools - ldirectord

To: Alois Treindl <alois@xxxxxxxx>, lvs-users@xxxxxxxxxxxxxxxxxxxxxx
Subject: Re: monitoring tools - ldirectord
From: andreas.koenig@xxxxxxxx (Andreas J. Koenig)
Date: 11 Jun 2001 10:26:04 +0200
>>>>> On Mon, 11 Jun 2001 09:47:45 +0200 (METDST), Alois Treindl 
>>>>> <alois@xxxxxxxx> said:

  > I think I will modify ldirectord so that when it tests a realserver
  > for service http, it also tests the mysql port on the dbms server,
  > and considers the http service as 'down' if the dbms is down.

  > I will try to make the patch to ldirectord reasonably general
  > and mail it to the maintainer (horms?)

I have doubts if patching ldirectord is the right approach. Please
consider more solutions that work independently of ldirectord. For
example:

1. Localizing your problem/defensive programming:

   Your webservers should be able to react sensitively to a database
   failure without relying on ldirectord anyway. Do they?

1.1 If the page they generate is OK for a user, there is no need to
    tell ldirectord anything about the problem

1.2 If they can generate a special page that says, for example
    "status=nodb" instead of "status=ok", you can simply test for
    "status=ok" in the ldirectord

In both cases your problem seems solved to me.

I'm not against patching ldirectord in general, but I'm unsure where
the line should be drawn between feature and bloat. Please, be
careful. There might be many more possible solutions if the above does
not work for you for some reason.

-- 
andreas


<Prev in Thread] Current Thread [Next in Thread>