Was this "out-of-sync" due to operations done via rpc? The option you
enabled is good, but not available of all database connectors. From
kamailio point of view, this can happen only when a new contact cannot
be stored due to unavailability of the database.
Cheers,
Daniel
On 16/01/2017 15:50, Vi
We did not purposely delete the records. But we were dealing with a setup
where the db was "out-of-sync" with memory. The reason I deleted them
during our tests was to replicate the issue we had in production.
The param we enabled fixed the issue within 30minutes by inserted when an
update failed.
You should not delete the record only from database via a db client
There is a rpc command to delete it, which will take care of removing it
from memory as well as from database (depending on the db_mode module
parameter for usrloc).
Cheers,
Daniel
On 16/01/2017 14:46, Vik Killa wrote:
> The rec
The record we added via RPC was first creating a new contact (and inserting
into the db), this was working fine. But we found that if we cleared the
database, any "updates" would fail. Adding that parameter caused the record
to get inserted if an update failed (re-register)
On Mon, Jan 16, 2017 a
The record you add via rpc is creating a new contact in memory or it's
updating an existing one?
Can you dump the record after you add it over rpc and send it over to
mailing list to see what attributes it has?
Cheers,
Daniel
On 14/01/2017 16:25, Vik Killa wrote:
> resolution update --
> we fo
resolution update --
we found that setting
`modparam("usrloc", "db_check_update", 1)`
fix the issue by inserting missing rows on re-reg
Thanks!
On Fri, Jan 13, 2017 at 9:30 AM, Vik Killa wrote:
> Hi Daniel,
> RPC flush is not setting the flag, but im not sure that is where the issue
> is, as
Hello,
I meant the rpc command to add contacts for aors, respectively the one
executed by:
/usr/local/sbin/kamcmd ul.add location ...
Not the ul.flush.
Cheers,
Daniel
On 13/01/2017 15:30, Vik Killa wrote:
> Hi Daniel,
> RPC flush is not setting the flag, but im not sure that is where the
> is
Hi Daniel,
RPC flush is not setting the flag, but im not sure that is where the issue
is, as I stated, we are not setting any memory-only flags with save()
But here is the flush function (FL_MEM not set)
static void ul_rpc_flush(rpc_t* rpc, void* ctx)
{
synchronize_all_udomains(0, 1);
return;
}
Hi,
We have tried using these flags:
save("location")
save("location", "0x00")
save("location", "0x04")
And still memory does not get flushed to DB.
I will test the RPC command.
Thanks,
/V
On Fri, Jan 13, 2017 at 9:12 AM, Daniel-Constantin Mierla wrote:
> Hello,
>
> that flag is used to mark
Hello,
that flag is used to mark a contact for storage only in memory. The
save() function has a parameter with flags where this kind of storage
can be set. Can you check the RPC command is setting this flag?
Cheers,
Daniel
On 13/01/2017 15:06, Vik Killa wrote:
> following up here
> i found if w
following up here
i found if we comment out a single line of code, kamcmd ul.flush works
here is the git diff
diff --git a/src/modules/usrloc/ucontact.c b/src/modules/usrloc/ucontact.c
index 47f3c2f..633ca81 100644
--- a/src/modules/usrloc/ucontact.c
+++ b/src/modules/usrloc/ucontact.c
@@ -474,
11 matches
Mail list logo