Re: syncd

To: " users mailing list." <lvs-users@xxxxxxxxxxxxxxxxxxxxxx>
Subject: Re: syncd
From: Roberto Nibali <ratz@xxxxxxxxxxxx>
Date: Sat, 25 Feb 2006 15:51:18 +0100
Matthias Krauss wrote:
i'm currently testing the "Server State Sync Demon", mcast is going well and established states are tranfered fine to the backup device, now i am looking for the persistent stats which are stated as NONE, those doesn't seem to get transfered to the backup device, i guess thats by design or is there a way to
change this ?

According to the code they should be "transfered" as well, reading

301 for (i=0; i<m->nr_conns; i++) {
302     unsigned flags;
304     s = (struct ip_vs_sync_conn *)p;
305     flags = ntohs(s->flags);
306     if (!(flags & IP_VS_CONN_F_TEMPLATE))
307             cp = ip_vs_conn_in_get(s->protocol,
308                     s->caddr, s->cport,
309                     s->vaddr, s->vport);
310     else
311             cp = ip_vs_ct_in_get(s->protocol,
312                     s->caddr, s->cport,
313                     s->vaddr, s->vport);
314     if (!cp) {
315             cp = ip_vs_conn_new(s->protocol,
316                     s->caddr, s->cport,
317                     s->vaddr, s->vport,
318                     s->daddr, s->dport,
319                     flags, NULL);
320             if (!cp) {
321                     IP_VS_ERR("ip_vs_conn_new failed\n");
322                     return;
323             }
324             cp->state = ntohs(s->state);

template connection entries (used for persistent connections) are created if they are not available already on the backup node.

Maybe I'm misreading it. I haven't used the synchronisation concept in years. Do you have any strange entries in your kernlog on either side?

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

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