Hello,
can you print the content of cls in frame 0 inside gdb?
p *cls
I found and fixed a potential issue, by now in master:
You can cherry pick it in branch 3.3 and try it, with:
git pull origin
git cherry-pick -x 4d15ba97bab58108cfedc45158d90583f67cdadf
But first get the content of cls, the fix I did was not convincing me it
is the reason for what you reported.
Cheers,
Daniel
On 8/23/12 9:29 AM, Daniel-Constantin Mierla wrote:
Hello,
On 8/22/12 6:34 PM, Øyvind Kolbu wrote:
On 2012-08-22 at 16:01, Daniel-Constantin Mierla
wrote:
Hello,
can you send the output of 'bt full' from gdb?
#0 0x002fd7da in
db_cluster_insert (_h=0xb7d2dc88, _k=0xbf9ac954,
_v=0xbf9ac7ec, _n=18) at dbcl_api.c:334
I will look over it when I get the first chance.
[...]
Also, can you give it a quick try with 9p instead
of 7p for both
connections?
9p for both works! At least not crashes.
Changed to:
modparam("db_cluster", "cluster",
"data2=>voip2_data2=9s9p;voip1_data2=9s9p")
modparam("usrloc", "db_mode", 3)
But it only writes to one of the databases. In this case the one
defined in voip1_data2.
Have you tried db_cluster together with the registrar-module?
I'd like all writing to go to both databases, but reading should be
failover-based.
I looked I my config used for testing during development and I
have an
use case with 9p/9p for two connections for write, tested with acc
module.
Cheers,
Daniel