<html>
<head>
<meta content="text/html; charset=ISO-8859-1"
http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix">On 9/12/12 3:06 PM, Gurjeet Singh
wrote:<br>
</div>
<blockquote
cite="mid:CABwTF4V+e7wteTjPn=65hmHNf9SWBuo=qna_nH_Z1r60XYVwxw@mail.gmail.com"
type="cite">
<div dir="ltr">
<div class="gmail_quote">On Wed, Sep 12, 2012 at 5:45 PM, Tatsuo
Ishii <span dir="ltr"><<a moz-do-not-send="true"
href="mailto:ishii@postgresql.org" target="_blank">ishii@postgresql.org</a>></span>
wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0
.8ex;border-left:1px #ccc solid;padding-left:1ex">
<div class="im">> So far I haven't heard of any damaging
or detrimental effects of -D option.<br>
> The only effect and overhead that I can verify is
that with -D option<br>
> pgpool will connect to every backend_hostname to
verify whether the<br>
> database backend is reachable. This may cause some
slowdown in initial<br>
> pgpool startup, but the upside is that the
backend_status value reflects<br>
> the true status of a backend.<br>
<br>
</div>
Really? At least Alex and I are not sure if it's a really
good idea,<br>
while no one is with your idea. Also I would like to hear
from Devrim<br>
since he is the mainter of init scripts.<br>
</blockquote>
<div><br>
Well, the only argument I have heard is that once added to
init script, it will be difficult to disable its effect,
like one could do by using $OPTS env variable. What I
consider 'damaging or detrimental' is if someone said
"That'll cause wrong results in certain mode of operation
(say replication)" or "this will cause a crash" or "this
will make some backend nodes unavailable via pgpool". If -D
option does not cause any unintended effects, I think it is
safe for me to use. I hope you get my point.<br>
</div>
</div>
</div>
</blockquote>
<br>
Safe for you to use doesn't justify changing the default behavior,
though. It just makes it safe for you to use.<br>
<br>
Consider the scenario in which someone disables a backend
intentionally to do some kind of work on it. In that case the user
would want the backend to remain disabled until the work is done and
he/she manually re-enables the backend. The current default acts
that way and users would reasonably expect v3.3 and above to keep
the same default behavior.<br>
<br>
Again, /etc/sysconfig/pgpool allows you to do whatever you want
--and shouldn't get overwritten by rpm updates or the spec file
needs fixing-- without changing expected behavior for everyone else.<br>
</body>
</html>