LVS
lvs-users
Google
 
Web LinuxVirtualServer.org

Re: Re: FOS seem not working normal?

To: Stephen Rowles <spr@xxxxxxxxxxxxxxx>
Subject: Re: Re: FOS seem not working normal?
Cc: lvs-users@xxxxxxxxxxxxxxxxxxxxxx
From: thomas.hoelsken@xxxxxx
Date: Fri, 20 Oct 2000 13:39:59 +0100


It seems, that my backup node tells the primary to shutdown the services for an
take-over; But the primary node reactivates himself and sends arp broadcast!

PRIMARY NODE

Oct 20 12:07:45 clu-lnx-01 pulse[25509]: gratuitous fos arps finished
Oct 20 12:07:45 clu-lnx-01 pulse[25218]: PARTNER HAS TOLD US TO GO INACTIVE!
Oct 20 12:07:45 clu-lnx-01 pulse[25518]: running command  "/sbin/ifconfig"
"eth0:1" "down"
Oct 20 12:07:45 clu-lnx-01 pulse[25218]: running command  "/usr/sbin/fos"
"--monitor" "-c" "/etc/lvs.cf" "--nofork"
Oct 20 12:07:45 clu-lnx-01 fos[25519]: Stopping local services (if any)
Oct 20 12:07:50 clu-lnx-01 pulse[25218]: fos process exited -- performing
service failover
Oct 20 12:07:50 clu-lnx-01 pulse[25218]: running command  "/usr/sbin/fos"
"--active" "-c" "/etc/lvs.cf" "--nofork"
Oct 20 12:07:50 clu-lnx-01 fos[25523]: Stopping local services (if any)
Oct 20 12:07:50 clu-lnx-01 pulse[25524]: running command  "/sbin/ifconfig"
"eth0:1" "192.168.126.65" "up"
Oct 20 12:07:50 clu-lnx-01 pulse[25522]: running command  "/usr/sbin/send_arp"
"-i" "eth0" "192.168.126.65" "0050DA1CBDC6" "192.168.126.255" "ffffffffffff"

BACKUP NODE

Oct 20 12:05:58 clu-lnx-02 pulse[6518]: gratuitous fos arps finished
Oct 20 12:05:58 clu-lnx-02 pulse[5272]: Notifying partner WE are taking control!
Oct 20 12:05:58 clu-lnx-02 pulse[5272]: fos process exited -- restarting it
Oct 20 12:05:58 clu-lnx-02 pulse[5272]: running command  "/usr/sbin/fos"
"--active" "-c" "/etc/lvs.cf" "--nofork"
Oct 20 12:05:58 clu-lnx-02 fos[6529]: Stopping local services (if any)
Oct 20 12:05:58 clu-lnx-02 pulse[6531]: running command  "/sbin/ifconfig"
"eth0:1" "192.168.126.65" "up"
Oct 20 12:05:58 clu-lnx-02 pulse[6530]: running command  "/usr/sbin/send_arp"
"-i" "eth0" "192.168.126.65" "005004D2713E" "192.168.126.255" "ffffffffffff"

Any suggestions why my backup wants to takeover and the primary don't give
control away!? This effect is written in the troubleshooting FAQ, but there is
no hint to get it work.

Many thanks,

Thomas Hoelsken
IT-TM-KS
C&N Touristic AG
Zimmersmühlenweg 55
61440 Oberursel
GERMANY
Phone: +49-6171-651198
Fax: +49-6171-6541198
Mail: thomas.hoelsken@xxxxxx



|--------+----------------------->
|        |          Stephen      |
|        |          Rowles       |
|        |          <spr@xxxxxxxx|
|        |          n.ac.uk>     |
|        |                       |
|        |          20.10.00     |
|        |          11:55        |
|        |                       |
|--------+----------------------->
  >----------------------------------------------------------------------------|
  |                                                                            |
  |       An:     lvs-users@xxxxxxxxxxxxxxxxxxxxxx                             |
  |       Kopie:  Thomas                                                       |
  |       Hölsken/Personal/VERWALTUNG/Neckermann_Reisen/DE@Neckermann_Reisen   |
  |       Thema:  Re: FOS seem not working normal?                             |
  >----------------------------------------------------------------------------|




At 06:39 20/10/2000 -0400, you wrote:
>On Fri, 20 Oct 2000 thomas.hoelsken@xxxxxx wrote:
>
> > The FTP-service is working, but the firtst connect takes nearly a minute.
>
>are you running ftpd under tcp wrappers? your inetd.conf will be something
>like
>
>ftp     stream  tcp     nowait  root    /usr/sbin/tcpd  wu.ftpd
>
>if so, you are having troubles with identd (lookup the HOWTO).
>change the line to
>
>ftp     stream  tcp     nowait  root    /usr/sbin/wu.ftpd  wu.ftpd

Alternatively you can tweak your hosts.allow, and you hosts file to allow
access.
make sure that the relevant machines in the cluster have local IP entries
in you /etc/hosts file, and are allowed in /etc/hosts.allow

The problem that I had was the inability to resolve host names for machines
within a NAT cluster because of their local only IP address. The minute
timeout is a "typical" name lookup timeout.

I don't know if this is entirely relevant.... I lost the previous setup
description e-mails. But I do know that disabling tcp wrappers is not
necessarily the right solution as it will disable logging and allowed IP
security checks. I have a NAT cluster running FTP and telnet and ssh, all
tcp wrappered with no access delays / problems.



>Joe
>
>--
>Joseph Mack mack@xxxxxxxxxxx
>
>

Steve.

----------------------------------------------------------------------------
Going to church doesn't make you a Christian any more than going to a garage
makes you a mechanic.







<Prev in Thread] Current Thread [Next in Thread>