Re: synchd

To: " users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: synchd
From: Roberto Nibali <ratz@xxxxxxxxxxxx>
Date: Sun, 26 Feb 2006 20:55:47 +0100
 >What would your expectation be regarding the synchronisation behaviour?
>We rarely got feedback from people using it, so the more input we get, the better can this behaviour be adapted.

with this setup the focus would be on the configured persistent stat taken over from the master to the backup.

I'm sorry, but I don't understand what you mean here. I'll read it again tomorrow morning.

>What do you mean by "VIP flagged down"? Of course you need to have the same config on each server, >anything else would result in promiscuousness regarding the receiving kernel side.

i was using syncd in combination with HA, the ip address takeover script from HA removes all VIPs in case of a outage of the master, i was testing this by alternating between master and backup and noticed that syncd works only properly if i readd the VIP (with status down) on the backup device after the ip address takeover to give
ipvsadm a chance to add the nodes.

Color my stupid but I have absolutely no idea what you mean by re-adding "the VIP with status down". Maybe you could send the output of

ip addr show; ip link show
ipvsadm -Ln
ipvsadm -Lnc

on both nodes, before and after the failover, so my simple mind can follow you.

Roberto Nibali, ratz
echo '[q]sa[ln0=aln256%Pln256/snlbx]sb3135071790101768542287578439snlbxq' | dc

<Prev in Thread] Current Thread [Next in Thread>
  • Re: synchd, Matthias Krauss
    • Re: synchd, Roberto Nibali <=