[OpenSIPS-Users] Long reload time for mi tls_reload for 200 tls/ssl certs

Bogdan-Andrei Iancu bogdan at opensips.org
Thu May 30 16:03:58 UTC 2024


Hi Denys,

That is really weird, 4 out of 5 traps point to the shm_malloc() 
function, trying to get a free mem check. What is the usage of the shm 
mem (use the shmem: stats class to see) ? How many times you do this 
reload? does it get slow from the first?

Regards,

Bogdan-Andrei Iancu

OpenSIPS Founder and Developer
   https://www.opensips-solutions.com
   https://www.siphub.com

On 14.05.2024 16:47, Denys Pozniak wrote:
> Hello!
>
> I disabled logging and added some resources to the virtual machine.
> On a working OpenSIPS, I reloaded the tls several times and in 
> parallel ran a trap.
> #opensips-cli -x mi tls_reload
> #opensips-cli -x trap
>
> If possible, please analyze it again, maybe you could find something 
> interesting:
> https://github.com/denyspozniak/opensips_tls_debug/tree/main
>
> Thanks in advance!
>
>
>
> ср, 8 мая 2024 г. в 19:59, Brett Nemeroff <brett at nemeroff.com>:
>
>     Just offering my experience here. I have, without a doubt,
>     noticed intensive logging brings a highly performant server to its
>     knees.
>
>     Disable ALL logging. Watch disk IO and confirm it's disabled. Try
>     it again. Just an easy thing to try.
>
>     -Brett
>
>
>     On Wed, May 8, 2024 at 7:17 AM Bogdan-Andrei Iancu
>     <bogdan at opensips.org> wrote:
>
>         Hi,
>
>         There is only one trap, ideally you should try to get several
>         during the reload time.
>
>         Still, the trap you did shows opensips doing some logging
>         (dumping to syslog) while reloading - could you check how
>         intensive this logging is and eventually to try to disable it
>         (increase the log level of opensips lower than INFO) to see if
>         there is any impact?
>
>         Regards,
>
>         Bogdan-Andrei Iancu
>
>         OpenSIPS Founder and Developer
>            https://www.opensips-solutions.com
>            https://www.siphub.com
>
>         On 08.05.2024 14:10, Denys Pozniak wrote:
>>         Hello!
>>
>>         If possible, please check log:
>>         https://github.com/denyspozniak/opensips_tls_debug/blob/main/gdb_opensips_20240508_115956
>>
>>
>>         ср, 8 мая 2024 г. в 08:55, Bogdan-Andrei Iancu
>>         <bogdan at opensips.org>:
>>
>>             Hi Denys.
>>
>>             That is rather weird, 250 certificates in 1 min. I assume
>>             it is not a DB issue (considering the db_text backend),
>>             so can you try to do multiple sequential "opensips-cli -x
>>             trap" to try to understand what is going on ?
>>
>>             Regards,
>>
>>             Bogdan-Andrei Iancu
>>
>>             OpenSIPS Founder and Developer
>>                https://www.opensips-solutions.com
>>                https://www.siphub.com
>>
>>             On 02.05.2024 11:41, Denys Pozniak wrote:
>>>             Hello!
>>>
>>>             There is a task to divide a single tls/ssl letsencrypt
>>>             certificate for white labels into specific ones.
>>>             I entered about ~250 certificates into db_text, but as
>>>             it turned out, for OpenSIPS this is a rather complex
>>>             operation to load them and takes about 1 minute and a
>>>             heavy CPU load is noticeable.
>>>
>>>             I would appreciate any advice on how to avoid this.
>>>
>>>             # wc -l dbtext/tls_mgm
>>>             253 dbtext/tls_mgm
>>>
>>>             # time opensips-cli -x mi tls_reload
>>>             "OK"
>>>             real 0m52.034s
>>>             user 0m1.419s
>>>             sys 0m0.433s
>>>
>>>             # time systemctl restart opensips
>>>             real    0m58.198s
>>>             user    0m0.024s
>>>             sys     0m0.055s
>>>
>>>             # opensips -V
>>>             version: opensips 3.4.4 (x86_64/linux)
>>>             flags: STATS: On, DISABLE_NAGLE, USE_MCAST, SHM_MMAP,
>>>             PKG_MALLOC, Q_MALLOC, F_MALLOC, HP_MALLOC, DBG_MALLOC,
>>>             FAST_LOCK-ADAPTIVE_WAIT
>>>             ADAPTIVE_WAIT_LOOPS=1024, MAX_RECV_BUFFER_SIZE 262144,
>>>             MAX_LISTEN 16, MAX_URI_SIZE 1024, BUF_SIZE 65535
>>>             poll method support: poll, epoll, sigio_rt, select.
>>>             git revision: 036d02961
>>>
>>>             -- 
>>>
>>>             BR,
>>>             Denys Pozniak
>>>
>>>
>>>
>>>             _______________________________________________
>>>             Users mailing list
>>>             Users at lists.opensips.org
>>>             http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>>
>>
>>
>>         -- 
>>
>>         BR,
>>         Denys Pozniak
>>
>>
>
>         _______________________________________________
>         Users mailing list
>         Users at lists.opensips.org
>         http://lists.opensips.org/cgi-bin/mailman/listinfo/users
>
>
>
> -- 
>
> BR,
> Denys Pozniak
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.opensips.org/pipermail/users/attachments/20240530/f0692b45/attachment.html>


More information about the Users mailing list