|
>After seeing Peter's posts on monitoring, I'm wondering if anyone
>has put together a blow by blow matrix/comparison chart of all
>the available monitoring methods (or at least the most well
>known or used ones) for lvs?
Thank you for volunteering :) ... what kind of information will you require?
offhand...
<most popular>
LVS + ldirectord + heartbeat + mon - It used to require mon due to "split
brain" bug, but the latest version should not require it. However, I
wouldn't personally choose this option without mon. There's a world of
features and "what-ifs" that you might get bit by without it. Note :
there's a complete rpm solution available from va-linux called ultramonkey
that simplifies setup quite a bit. I don't know how out of date it is now..
LVS + mon + heartbeat - very customizable, good setup overall. this one has
stood the test of time. I bet it's harder to setup for new people.
<newest technology>
LVS + VRRP (+mon?) - active/active clusters supported, technology is
specifically designed with LVS in mind (after adoption from Cisco). I'm not
sure what kind of production status VRRP is at, but it looks like it has
stabled out.
|
