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

List:       keepalived-devel
Subject:    Re: [Keepalived-devel] Existing connections if TCP_CHECK fails?
From:       Todd Fleisher <todd () fleish ! org>
Date:       2010-09-16 19:56:22
Message-ID: A5738A5B-5B8C-468C-8572-E957C41797F9 () fleish ! org
[Download RAW message or body]

AFAIK, once a connection is established - a failed health check won't impact it in \
any way. Now if there was a failed network path between the client & server that \
caused that caused the health check to fail - then you might also see an interruption \
of existing connections - but it wouldn't be that the failed health check would have \
caused that - the failed network path would have.

-T

On Sep 16, 2010, at 12:21 PM, Paul Hirose wrote:

> I'm getting a bit of flapping on my health-checks to my realservers.  My delay-loop \
> is 5 and I'm using TCP_CHECK on keepalived 1.1.20 on RHEL 5.5 64bit.  In this case \
> it's port 22/TCP, SSH with lb_kind DR. 
> About once or twice a day, my health-check fails and I get an email about it.  \
> Usually, the next health-check 5sec later, brings the service back up.  During \
> those 5 secs, what happens to existing connections?  If the connection is idle, I \
> suppose the a couple of SSH keepalive checks will recover the connection, or it \
> might not ever even notice keepalived flagged the realserver is down?  But if there \
> happens to be traffic flowing through my SSH session and during that time, a \
> realserver is flagged as down and then back up delay-loop secs later, what happens \
> to the transmission/data during those 5 secs?  It could, I suppose, just rely on \
> TCP retries but I rather than my guessing, I thought I'd just ask how keepalived \
> (and/or IPVS?) handles this.  Doesn't keepalived actually instruct ipvs to remove \
> the realserver for those 5 secs? 
> I figure new connections during the failed health-check time simply go to one of my \
> other realservers that are still flagged as up, so I'm not worried about new \
> connections. 
> And yes, the real issue is why is the SSH server not accepting connections at that \
> exact moment when keepalived tried a health-check.  But that's a different issue, \
> and I'm not sure I have much control over the SSH server. 
> Thank you,
> PH
> PS
> As a feature request for near-term, I'd love to see the nb_retries type of thing \
> enabled for TCP_CHECK and/or MISC_CHECK 
> ==
> Paul Hirose
> University of California, Davis
> pthirose@ucdavis.edu
> ------------------------------------------------------------------------------
> Start uncovering the many advantages of virtual appliances
> and start using them to simplify application deployment and
> accelerate your shift to cloud computing.
> http://p.sf.net/sfu/novell-sfdev2dev
> _______________________________________________
> Keepalived-devel mailing list
> Keepalived-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/keepalived-devel
> 
> -- 
> This message has been scanned for viruses and
> dangerous content, and is believed to be clean.
> 


------------------------------------------------------------------------------
Start uncovering the many advantages of virtual appliances
and start using them to simplify application deployment and
accelerate your shift to cloud computing.
http://p.sf.net/sfu/novell-sfdev2dev
_______________________________________________
Keepalived-devel mailing list
Keepalived-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/keepalived-devel


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