Mailing List Archive

Help - Dead Director
Help please - I am having some serious problems with sick director box.
I
had the box in service for a long time and moved it last week. When I
changed the IPs it started behaving poorly.
It is a simple NAT box and is not even load-balancing currently - just
port-forwarding really. Kernel 2.2.16 with all ipvs up-to-date.

Symptoms are:
- Seems to drop the virtual host IP's that it listens on. ifconfig shows
them all there but when I hit it with traffic for the ones that seem
down
down I
don't see anything; either through the logging of denied packets
(ipchains) in my kernel messages or through tcpdump
- when I turn the virtual IP that is broke over to the main IP then back
again the virtual device seems to come up - my problem is that it will
fall over again.
- nothing in my kernel logs, ifconfig shows healthy interfaces, and
netstat is all ok
- tcpdump outputs seem slow sometimes
- sometimes when I ask for "route" it is really slow
- ipvsadm -l is extremely slow to show output - all the time

Seems like the kernel is all bound up doing something. Reboots don't
help.

help please, I know this is a little off-topic. If I should post to
another list I am open to suggestions. I haven't found much help in
faq's
etc and time is of essence.

thank you!
-TRevor

ps - this is a resend as I didn't know this was a members-only list and
sent from my other address. oops.
Help - dead director [ In reply to ]
Help please - I am having some serious problems with sick director box. I
had the box in service for a long time and moved it last week. When I
changed the IPs it started behaving poorly.
It is a simple NAT box and is not even load-balancing currently - just
port-forwarding really. Kernel 2.2.16 with all ipvs up-to-date.

Symptoms are:
- Seems to drop the virtual host IP's that it listens on. ifconfig shows
them all there but when I hit it with traffic for the ones that seem down
down I
don't see anything; either through the logging of denied packets
(ipchains) in my kernel messages or through tcpdump
- when I turn the virtual IP that is broke over to the main IP then back
again the virtual device seems to come up - my problem is that it will
fall over again.
- nothing in my kernel logs, ifconfig shows healthy interfaces, and
netstat is all ok
- tcpdump outputs seem slow sometimes
- sometimes when I ask for "route" it is really slow
- ipvsadm -l is extremely slow to show output - all the time

Seems like the kernel is all bound up doing something. Reboots don't help.

help please, I know this is a little off-topic. If I should post to
another list I am open to suggestions. I haven't found much help in faq's
etc and time is of essence.

thank you!
-TRevor
Re: Help - Dead Director [ In reply to ]
Everything you are describing sounds like DNS name resolution problems.

Use a "-n" to the commands you are trying to tell them not to bother
resolving IP addresses back to host names.

You may not even need DNS on your servers. Try naming /etc/resolv.conf out
for a moment and see if that makes things run smooth.

--K

>From: Trevor Marshall <trevor@risk2risk.com>
>Reply-To: lvs-users@LinuxVirtualServer.org
>To: Linux Virtual Server Mail List <lvs-users@LinuxVirtualServer.org>
>Subject: Help - Dead Director
>Date: Tue, 21 Nov 2000 13:01:59 +0000
>
>Help please - I am having some serious problems with sick director box.
>I
>had the box in service for a long time and moved it last week. When I
>changed the IPs it started behaving poorly.
>It is a simple NAT box and is not even load-balancing currently - just
>port-forwarding really. Kernel 2.2.16 with all ipvs up-to-date.
>
>Symptoms are:
>- Seems to drop the virtual host IP's that it listens on. ifconfig shows
>them all there but when I hit it with traffic for the ones that seem
>down
>down I
>don't see anything; either through the logging of denied packets
>(ipchains) in my kernel messages or through tcpdump
>- when I turn the virtual IP that is broke over to the main IP then back
>again the virtual device seems to come up - my problem is that it will
>fall over again.
>- nothing in my kernel logs, ifconfig shows healthy interfaces, and
>netstat is all ok
>- tcpdump outputs seem slow sometimes
>- sometimes when I ask for "route" it is really slow
>- ipvsadm -l is extremely slow to show output - all the time
>
>Seems like the kernel is all bound up doing something. Reboots don't
>help.
>
>help please, I know this is a little off-topic. If I should post to
>another list I am open to suggestions. I haven't found much help in
>faq's
>etc and time is of essence.
>
>thank you!
>-TRevor
>
>ps - this is a resend as I didn't know this was a members-only list and
>sent from my other address. oops.
>
>_______________________________________________
>LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
>Send requests to lvs-users-request@LinuxVirtualServer.org
>or go to http://www.in-addr.de/mailman/listinfo/lvs-users

_____________________________________________________________________________________
Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com
Re: Help - dead director [ In reply to ]
Sorry if this posts twice..

Sounds like you are having DNS trouble. Shut off DNS by naming out
/etc/resolv.conf (no need to reboot).

Also, try using the commands with "-n" options to tell them not to convert
IP addresses back in to host names.

--K

>From: Trevor Marshall <pmarshal@uoguelph.ca>
>Reply-To: lvs-users@LinuxVirtualServer.org
>To: lvs-users@linuxvirtualserver.org
>Subject: Help - dead director
>Date: Tue, 21 Nov 2000 07:51:15 -0500 (EST)
>
>Help please - I am having some serious problems with sick director box. I
>had the box in service for a long time and moved it last week. When I
>changed the IPs it started behaving poorly.
>It is a simple NAT box and is not even load-balancing currently - just
>port-forwarding really. Kernel 2.2.16 with all ipvs up-to-date.
>
>Symptoms are:
>- Seems to drop the virtual host IP's that it listens on. ifconfig shows
>them all there but when I hit it with traffic for the ones that seem down
>down I
>don't see anything; either through the logging of denied packets
>(ipchains) in my kernel messages or through tcpdump
>- when I turn the virtual IP that is broke over to the main IP then back
>again the virtual device seems to come up - my problem is that it will
>fall over again.
>- nothing in my kernel logs, ifconfig shows healthy interfaces, and
>netstat is all ok
>- tcpdump outputs seem slow sometimes
>- sometimes when I ask for "route" it is really slow
>- ipvsadm -l is extremely slow to show output - all the time
>
>Seems like the kernel is all bound up doing something. Reboots don't help.
>
>help please, I know this is a little off-topic. If I should post to
>another list I am open to suggestions. I haven't found much help in faq's
>etc and time is of essence.
>
>thank you!
>-TRevor
>
>
>_______________________________________________
>LinuxVirtualServer.org mailing list - lvs-users@LinuxVirtualServer.org
>Send requests to lvs-users-request@LinuxVirtualServer.org
>or go to http://www.in-addr.de/mailman/listinfo/lvs-users

_____________________________________________________________________________________
Get more from the Web. FREE MSN Explorer download : http://explorer.msn.com