On Fri, May 25, 2007 at 10:47:56AM +0900, Horms wrote:
> On Thu, May 24, 2007 at 04:51:13PM -0700, Joseph Mack NA3T wrote:
> > On Thu, 24 May 2007, Gerry Reno wrote:
> >
> > >I'm getting a lot of these on various links/tabs on the
> > >http://www.linuxvir
Documents.html on this server.
Sorry about this. I'm actually not sure what the problem is.
But I am investigating. Hopefully it will be fixed shortly.
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/
___
Linux
se proc values
were removed many moons ago.
In relation to the actual problem. After you make this change,
what to the timeouts for (new) connections in the established
state look like. You can see them by examining ipvsadm -Lcn
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.
elf, mainly) that the list
> delivery is optimised.
Seems to be workign fine here.
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/
___
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requ
timeout, rather than full fledged connection entries.
>
> What's the difference?
>
> and at LVS2, when I connecting using F5(reload),
> I can see over 600 connections, but LVS1 only 5-8 connections.
>
> Is
etty easy to debug using tcpump -ne
or similar.
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/
___
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to [EMAIL PROTECTED]
or go to http://
ld add another rs, or change the
> weight of an existing rs, all works as intended.
>
> I have tried to first set the weight to 0, and then remove it, with the same
> result.
Thanks, I'll look into that.
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co
x27;Quiescent' instead of 'Restored'?
Thanks, I'll look into getting that fixed.
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/
___
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServ
this case that would be somewhat silly.
If you want to use LVS for this, wouldn't an easy way be
to bind the processes to 127.0.0.1 and 127.0.0.2 respectively
and set them up as the real-servers in LVS.
That said, using fork-on-connect or preforking in the user-space
application, and making s
e a problem for you, then perhaps you will benifit from
splitting things up a bit. Though to be honest, how you do this is
really its just a matter of personal taste.
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/
_
nSwap master status I
> always get "master stopped". I believe my configuration on the master and
> backup machine is fine even the lo and iproute on the real server.
> So what is it I am doing run. Here is the version of the LVS I am using:
> LVSSyncDaemonSwap.in,v 1.7 2006/05/1
ering if Graeme's solution might not break the persistancy
that you are after. It seems to me that by breaking up the virtual
service and load balancing in groups based on the source-port, that
there is nothing to stop a given client coming from different ports
(in different ranges) for the s
be randomised to mitigate certain classes of security threats.
--
Horms
H: http://www.vergenet.net/~horms/
W: http://www.valinux.co.jp/en/
___
LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
Send requests to [EMAIL PROTECTED]
intermediate copy, then
ldirectord would attempt to load it. This is compounded by the problem
that regardless of what causes ldirectord to reload its config, if the
config is invalid then ldirectord will exit with an error.
I guess that implementing ldirectord checkconfig would be a good idea.
But
and if i would add another rs, or change the
> weight of an existing rs, all works as intended.
>
> I have tried to first set the weight to 0, and then remove it, with the same
> result.
Hi,
I finally found some time to look into this problem.
Yes, it does look like it is somewhat
ds,
Hi,
I finally found some time to look into this problem.
Yes, it does look like it is somewhat broken.
I have made two patches which I believe solve this problem,
I will send them as "[PATCH] [LDIRECTORD] Fix logic bug in _status_down"
and "[PATCH] [LDIECTORD] Remove stale entries on
Fix a logig bug in _status_down() that prevented fallback servers
from being removed if
a) at start no other real servers were present and
b) a real server was subsequently added
The new logic is the logical not of that found in _status_up
With thanks to Sebastian Vieira
Signed-off-by: Simon Ho
Call ld_start() on reload to flush out stale entires.
With thanks to Sebastian Vieira
Signed-off-by: Simon Horman <[EMAIL PROTECTED]>
Index: heartbeat/ldirectord/ldirectord.in
===
--- heartbeat.orig/ldirectord/ldirectord.in 2007
Masq1024 0 0
Hi Struan,
Sorry for letting your bug report fester in my inbox for such a long
time. I've been doing a bit of a blitz on ldirectord bugs over the past
few days, and I found that inadvertantly I had a fix for this problem -
which is still present in the l
On Wed, Jul 04, 2007 at 12:02:10PM +0200, Sebastian Vieira wrote:
> On 7/4/07, Horms <[EMAIL PROTECTED]> wrote:
> >
> > I finally found some time to look into this problem.
> > Yes, it does look like it is somewhat broken.
> > Can you see if the attached patch
20 matches
Mail list logo