[Freeswitch-users] Strange "timer_test" result

[Christopher Rienzo]

The crash in the old posix timer is actually caused by a glibc
timer_create() race condition when using SIGEV_THREAD.  So, you could
should be able to use the old timer if you patch glibc.  I figured that was
not a good solution for end users, so I switched to SIGEV_SIGNAL which did
degrade performance.

Chris



On Wed, Jun 27, 2012 at 4:44 AM, Stephen Wilde <wstephen80 at gmail.com> wrote:

> My experience in FS timers:
>
> 1. voice quality is related to timer_test result so if you have irregular
> result, you'll have bad voice quality (and low ACD)
>
> 2. mod_posix_timer is my solution (I haven't try timerfd because I'm using
> CentOS 5.8).
>
> 3. there are two version of mod_posix_timer, the "old" one (the one before
> the fix of git 'FS-4064 mod_posix_timer workaround of CentOS 5 glibc
> problem'  of 30 april) and the 'new' one (present in latest git)
>
> 4. the 'old' mod_posix_timer is much more efficient that 'new' one (with
> 'new' mod_posix_timer is half way between 'old' mod_posix_timer and soft
> timer).
>
> 5. loading the 'old' mod_posix_timer there are random FS crashes, also if
> you don't use the posix timer in your profile
>
> 6. the 'new' mod_posix_timer solves the crashes of FS
>
> So my conclusion is that the 'old' mod_posix_timer is the best one but
> sometimes FS crashes but the 'new' mod_posix_timer is not good as the 'old'
> one but it solves the crash issue.
>
> In my production system I'm using the 'old' one in servers with high load
> and the 'new' one in serves with less load, according with timer_test
> results.
>
> Memory leak: I have done some test mixing different versions of FS and
> mod_posix_timer and in any case, if I use the latest Freeswitch git the
> used memory grows up continously, also if I use 'old' mod_posix_timer so,
> my conclusion is that the memory leakage is related to Freeswitch. In my
> production server, latest FS git + old mod_posix_timer consumes all server
> memory (12MB) in 2 days. For this reason, in all my installation I'm using
> the Freeswitch git of 26th march with new or old mod_posix_timer (depending
> on timer_test result).
>
> Stephen
>
> On Wed, Jun 27, 2012 at 12:00 AM, marksly <mark at callassistants.com> wrote:
>
>> I am having the same issue as Stephen Wilde regarding the strange
>> "timer_test" results. Was it determined that the mod_posix timer is the
>> best
>> solution? Likewise with Stephen, any timer except 20ms works great.
>>
>> --
>> View this message in context:
>> http://freeswitch-users.2379917.n2.nabble.com/Strange-timer-test-result-tp7399299p7580297.html
>> Sent from the freeswitch-users mailing list archive at Nabble.com.
>>
>> _________________________________________________________________________
>> Professional FreeSWITCH Consulting Services:
>> consulting at freeswitch.org
>> http://www.freeswitchsolutions.com
>>
>> 
>> 
>>
>> Official FreeSWITCH Sites
>> http://www.freeswitch.org
>> http://wiki.freeswitch.org
>> http://www.cluecon.com
>>
>> Join Us At ClueCon - Aug 7-9, 2012
>>
>> FreeSWITCH-users mailing list
>> FreeSWITCH-users at lists.freeswitch.org
>> http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
>> UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
>> http://www.freeswitch.org
>>
>
>
> _________________________________________________________________________
> Professional FreeSWITCH Consulting Services:
> consulting at freeswitch.org
> http://www.freeswitchsolutions.com
>
> 
> 
>
> Official FreeSWITCH Sites
> http://www.freeswitch.org
> http://wiki.freeswitch.org
> http://www.cluecon.com
>
> Join Us At ClueCon - Aug 7-9, 2012
>
> FreeSWITCH-users mailing list
> FreeSWITCH-users at lists.freeswitch.org
> http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
> UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users
> http://www.freeswitch.org
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://lists.freeswitch.org/pipermail/freeswitch-users/attachments/20120627/712ca5ca/attachment.html