[OpenSIPS-Users] memory management and drouting

Noel R. Morais noelrocha at gmail.com
Thu Jun 11 18:30:30 CEST 2009


Hi Guys,

I'm really interested in this thread.

few months ago I was planing to use drouting to solve my "number
portability" problem. The problem is that we are planing to have
something near 15 millions of routes (today it is only 1 million, but
it is growing), all of the routes are prefixed with full numbers, what
I mean: country code + area code + phone number.

What would be the estimated amount of memory to handle this crazy
number of rules?

Thanks,

Noel

On Thu, Jun 11, 2009 at 12:10 PM, Brett Nemeroff<brett at nemeroff.com> wrote:
> Bogdan,
> In the interest of the mailing list archives....
> I increased my server memory from 2G to 4G and increased the SHM using the
> -m param on the cmd line to 2048 (!!). And the problems seem to have gone
> away..  opensipsctl fifo dr_reload does take about a minute to complete. It
> would be fantastic to see in the server logs "dr_reload complete!" :D
> I thikn it was just suffering from serious memory starvation. Question.. are
> the route tables compiled out (dr_rules and dr_gateways) or are they
> seperate memory lookups (ie: if I add a dr_gateway, does my 1 million row
> table now become 2 million?)
> Thanks,
> Brett
>
> On Thu, Jun 11, 2009 at 2:50 AM, Bogdan-Andrei Iancu
> <bogdan at voice-system.ro> wrote:
>>
>> Hi Brett,
>>
>> What it happens is the attendant process spending too much time in doing
>> shutdown - this trigger the CRITICAL message you see after 60 seconds. So if
>> the shutdown is longer than 60 seconds, opensips will brutally stop with
>> this error.
>>
>> Why it happens ? my guess is it takes longer than 60 secs to free from mem
>> all the 1M of rules from drouting :)....This is the optimist guess. The
>> pessimist one is maybe some end looping (bug?) happens and opensips never
>> stops.....
>>
>> Regards,
>> Bogdan
>>
>> Brett Nemeroff wrote:
>>>
>>> This may be a hint?
>>> Jun  9 09:32:31 voicefoxtelephony /usr/local/sbin/opensips[5358]:
>>> INFO:core:sig_usr: signal 15 received
>>> Jun  9 09:32:31 voicefoxtelephony /usr/local/sbin/opensips[5340]:
>>> INFO:core:sig_usr: signal 15 received
>>> Jun  9 09:32:31 voicefoxtelephony /usr/local/sbin/opensips[5338]:
>>> INFO:core:sig_usr: signal 15 received
>>> Jun  9 09:32:57 voicefoxtelephony ntpd[2123]: synchronized to
>>> 64.73.32.134, stratum 2
>>> Jun  9 09:33:31 voicefoxtelephony /usr/local/sbin/opensips[5326]:
>>> CRITICAL:core:sig_alarm_abort: BUG - shutdown timeout triggered, dying...
>>>
>>> I stopped the proc with opensipsctl stop
>>>
>>>
>>> On Tue, Jun 9, 2009 at 9:31 AM, Brett Nemeroff <brett at nemeroff.com
>>> <mailto:brett at nemeroff.com>> wrote:
>>>
>>>    This has actually just become a pretty serious issue for me....
>>>
>>>    It seems that now when I restart opensips with opensipsctl
>>>    stop/start one process hangs:
>>>     ps ax|grep opensips
>>>     4744 ?        R      0:57 /usr/local/sbin/opensips -P
>>>    /var/run/opensips.pid -m 512
>>>
>>>    Which will eventually die in like 2-3 minutes.. but then I have to
>>>    delete /var/run/opensips.pid
>>>
>>>    Also, as of late, I'm getting the error that my dr_rules table is
>>>    empty, which it certainly isn't..This all seems to be related to
>>>    memory issues.. any ideas here??
>>>    -Brett
>>>
>>>
>>>
>>>
>>>    On Mon, Jun 8, 2009 at 9:15 PM, Brett Nemeroff <brett at nemeroff.com
>>>    <mailto:brett at nemeroff.com>> wrote:
>>>
>>>        management
>>>
>>>
>>>
>>> ------------------------------------------------------------------------
>>>
>>> _______________________________________________
>>> Users mailing list
>>> Users at lists.opensips.org
>>> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>>
>>
>
>
> _______________________________________________
> Users mailing list
> Users at lists.opensips.org
> http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>



More information about the Users mailing list