Hi Henning/Daniel,
Sorry to take such a long time replying back.
For my rig, I am using the memory as location back-up. When the issue
occurs, I see many registration request sent from SIPp but not many 200
replies. netstat shows a huge amount of packets get dropped.
Also I realized the issue only occurs when SIPp tried to register many users
in a very short time. With the same number of users, the issue goes away if
registration rate is kept lower. When the issue occurs, usrloc module
contends lock a lot calling from new_ucontact(), and many system time is
spent in yield system calls. So it looks more like a performance issue to
me.
Thanks,
-Joy
On 1/27/09, Henning Westerholt <henning.westerholt(a)1und1.de> wrote:
On Monday 26 January 2009, joy yue wrote:
Is there a limitation on the number of rows in
location table? In my rig,
whatever number of users I use (>2million users), I notice the number of
users in location table is 343707, which is far less than the number of
users in my test. I thought previously that location table saves all the
users in my test.
Also when I use large number of users (>2million), openser pops up an
error: ERROR: registrar:update_contacts: invalid cseq for aor <xxxx>. Has
anyone saw this before? I am using openser1.3.2.
Hi joy,
no, there is no such a limitiation, we've more registered users. The
invalid
CSEQ error you see is not related to this observation. The error means that
a
device tried to do re-registration (same callid), but without properly
increasing the Cseq number in the REGISTER request - RFC3261 says that
requests from the same dialog (like REGISTER + re-REGISTER) must have
increasing cseq.
What db_mode do you use in your usrloc? Do you see any other errors in the
logs?
Cheers,
Henning