[Serusers] Userloc replication via FIFO

[Daryl Sanders]

Thanks for all the help guys! 
The initial problem was how to deal with replication of contacts
modified via the web. However this has made me re-think my setup. This
is what I've come up with so far. I would really appreciate it if you
could let me know of any pitfalls you might see.

It looks like my best option is to execute sipsak from php within my
web interface to add/delete contacts, this way they can go through the
normal call processing of ser.cfg and be replicated via t_replicate or
forward_tcp. I will have the same web interface & mysql running on
both boxes... it's just a band-aid until xmlrpc in SER is mature, then
I'll move web & mysql to dedicated boxes. I haven't the time to modify
LVS right now so I will most likely use VRRP in a stand-by HA setup
with a single IP. I am really curious about LVS_NAT if anyone has any
experiences to share. A review I read shows it performing well on up
to 2000 requests per/sec.

I am planning to use save_memory on SER replications, and do 2-way
MySQL replication of everything on the back-end. I have seen Paul's
patch for preventing save_memory from writing to MySQL, and I assume I
will want to use this to make sure back-end DB replication is handled
by MySQL only.

Right now I am looking for stability. I realize this may not scale
well until the LVS & xmlrpc pieces are in place, but maybe it will buy
me some time.

Thanks again for your valuable input!

- Daryl





On 7/10/05, Greger V. Teigre <greger at teigre.com> wrote:
> Daryl,
> See my recent post to Jack Wei on the underlying issues related to
> replication and the use of save_memory().
> g-)
> 
> Daryl Sanders wrote:
> > I'm looking for a way to replicate userloc changes to another SER box
> > via FIFO. I am using FIFO via PHP to manipulate contacts, but I'm not
> > sure how to replicate those changes to my other SER's memory.  Any
> > thoughts? Any help would be greatly appreciated. I'm kinda stumped on
> > this one.
> >
> > - Daryl
> >
> > _______________________________________________
> > Serusers mailing list
> > serusers at lists.iptel.org
> > http://lists.iptel.org/mailman/listinfo/serusers
> 
>