Different versions, different cpu load
Hello! I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider. Regards
No one experiment this same issue? I also noticed a high cpu consume on call hangup. Regards 2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk. 2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable "We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower "I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower "Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
sounds like a good opportunity to try git-bisect to locate where the performance change occurs. On Thu, Aug 5, 2010 at 10:37 AM, Eliot Gable <egable+freeswitch@gmail.com<egable%2Bfreeswitch@gmail.com>
wrote:
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems
with
odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- -Rupa
ubunto is a nice linux distrib for audio station, not servers... ----- Original Message ----- From: "Brian West" <brian@freeswitch.org> To: "FreeSWITCH Users Help" <freeswitch-users@lists.freeswitch.org> Sent: Thursday, August 05, 2010 2:50 PM Subject: Re: [Freeswitch-users] Different versions, different cpu load
Or stop using ubuntu.
/b
On Aug 5, 2010, at 12:58 PM, Rupa Schomaker wrote:
sounds like a good opportunity to try git-bisect to locate where the performance change occurs.
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
Ubuntu is a common theme when someone says "OMG IT USES ALL MY CPU!" /b On Aug 5, 2010, at 2:09 PM, Madovsky wrote:
ubunto is a nice linux distrib for audio station, not servers...
----- Original Message ----- From: "Brian West" <brian@freeswitch.org> To: "FreeSWITCH Users Help" <freeswitch-users@lists.freeswitch.org> Sent: Thursday, August 05, 2010 2:50 PM Subject: Re: [Freeswitch-users] Different versions, different cpu load
Or stop using ubuntu.
/b
Which distro do you sugest? Regards 2010/8/5 Brian West <brian@freeswitch.org>
Ubuntu is a common theme when someone says "OMG IT USES ALL MY CPU!"
/b
On Aug 5, 2010, at 2:09 PM, Madovsky wrote:
ubunto is a nice linux distrib for audio station, not servers...
----- Original Message ----- From: "Brian West" <brian@freeswitch.org> To: "FreeSWITCH Users Help" <freeswitch-users@lists.freeswitch.org> Sent: Thursday, August 05, 2010 2:50 PM Subject: Re: [Freeswitch-users] Different versions, different cpu load
Or stop using ubuntu.
/b
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
I thinkg it is very difficult to find it without a good knowledge of FS internals. Do you know if there is any way to make a profiling which could help? Regards 2010/8/5 Rupa Schomaker <rupa@rupa.com>
sounds like a good opportunity to try git-bisect to locate where the performance change occurs.
On Thu, Aug 5, 2010 at 10:37 AM, Eliot Gable <egable+freeswitch@gmail.com<egable%2Bfreeswitch@gmail.com>
wrote:
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems
with
odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- -Rupa
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
Not really. If you have a test that shows "good/bad" we (devs) can find the reason for a problem. git handles choosing versions for testing and you handle the "good/bad" decision. The end result is "here is the git version where performance changes" which tony or whomever can then look at the patch and decide what he wants to do about it. 2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
I thinkg it is very difficult to find it without a good knowledge of FS internals. Do you know if there is any way to make a profiling which could help?
Regards
2010/8/5 Rupa Schomaker <rupa@rupa.com>
sounds like a good opportunity to try git-bisect to locate where the performance change occurs.
On Thu, Aug 5, 2010 at 10:37 AM, Eliot Gable <egable+freeswitch@gmail.com<egable%2Bfreeswitch@gmail.com>
wrote:
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some
weeks
ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- -Rupa
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- -Rupa
Latest revisions seem to solve this issue. I'll continue with the tests... 2010/8/6 Rupa Schomaker <rupa@rupa.com>
Not really. If you have a test that shows "good/bad" we (devs) can find the reason for a problem. git handles choosing versions for testing and you handle the "good/bad" decision. The end result is "here is the git version where performance changes" which tony or whomever can then look at the patch and decide what he wants to do about it.
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
I thinkg it is very difficult to find it without a good knowledge of FS
internals. Do you know if there is any way to make a profiling which could help?
Regards
2010/8/5 Rupa Schomaker <rupa@rupa.com>
sounds like a good opportunity to try git-bisect to locate where the performance change occurs.
On Thu, Aug 5, 2010 at 10:37 AM, Eliot Gable < egable+freeswitch@gmail.com <egable%2Bfreeswitch@gmail.com>> wrote:
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some
weeks
ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- -Rupa
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- -Rupa
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
I've got about 75% decrease in performarce. I am using same server and same config. DB was already in ramdisk as svn version so I don't find an explanation for it... Regards 2010/8/5 Eliot Gable <egable+freeswitch@gmail.com<egable%2Bfreeswitch@gmail.com>
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems
with
odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
Were both of your kernel running on 1000HZ ? 2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
I've got about 75% decrease in performarce. I am using same server and same config. DB was already in ramdisk as svn version so I don't find an explanation for it...
Regards
2010/8/5 Eliot Gable <egable+freeswitch@gmail.com>
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
How do I tell if the kernel is running at 1000hz 2010-8-6 下午4:29在 "Seven Du" <dujinfang@gmail.com> 提到： Were both of your kernel running on 1000HZ ? 2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
I've got about 75% decrease in performarce. I am using same server and same config. DB was alrea... Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@list...
These compariso is done into the same machine. I run SVN version, make the test, stop SVN version, run GIT version and make the test again. It runs over a "Intel(R) Core(TM)2 Duo CPU E7500 @ 2.93GHz" Regards 2010/8/6 Seven Du <dujinfang@gmail.com>
Were both of your kernel running on 1000HZ ?
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
I've got about 75% decrease in performarce. I am using same server and same config. DB was already in ramdisk as svn version so I don't find an explanation for it...
Regards
2010/8/5 Eliot Gable <egable+freeswitch@gmail.com<egable%2Bfreeswitch@gmail.com>
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some
weeks
ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
The default Ubuntu kernel running on 100Hz/250Hz. And there might be some great changed in timing between your two versions. I don't remember exactly. And I don't know if it's the reason causing load. I use the following command to see Hz, it should approximately increase 100 or 250 or 1000 each second. watch grep interr /proc/interrupts Also, can you try start FS with -nocal to see the difference? I still using Ubuntu because it need to schedule down time to migrate to centos in our datacenter. So I really would like to see it works ok on Ubuntu. 2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
These compariso is done into the same machine. I run SVN version, make the test, stop SVN version, run GIT version and make the test again.
It runs over a "Intel(R) Core(TM)2 Duo CPU E7500 @ 2.93GHz"
Regards
2010/8/6 Seven Du <dujinfang@gmail.com>
Were both of your kernel running on 1000HZ ?
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
I've got about 75% decrease in performarce. I am using same server and same config. DB was already in ramdisk as svn version so I don't find an explanation for it...
Regards
2010/8/5 Eliot Gable <egable+freeswitch@gmail.com>
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>
Hello!
I was been some test using one FS tarball version downloaded some weeks ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some problems with odbc connections managemens. Now I'm using a git version (FreeSWITCH Version 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to solve this problem but I noticed it has a high cpu load comparing with svn version. While I could manage more than 200 calls with a 75% CPU load into a dual core server using svn version, now, 50 calls consume this 75% cpu. I can see same modules are loaded (except new hash module needed in git version) and same scenario is used. I am not be able to find why now it uses more cpu than before. Any idea? Each testing call is a simple bridge to an external sip provider.
Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
I cannot see updates/seconds using that command. I can see following boot params for kernel at /boot/config-2.6.32-21-server: CONFIG_NO_HZ=y CONFIG_HZ_100=y # CONFIG_HZ_250 is not set # CONFIG_HZ_300 is not set # CONFIG_HZ_1000 is not set CONFIG_HZ=100 I have been reading about this task and people recomends using 1000 HZ instead 100 ¿what do you think? Centos seems use 1000 by default. Regards 2010/8/6 Seven Du <dujinfang@gmail.com>
The default Ubuntu kernel running on 100Hz/250Hz. And there might be some great changed in timing between your two versions. I don't remember exactly. And I don't know if it's the reason causing load.
I use the following command to see Hz, it should approximately increase 100 or 250 or 1000 each second.
watch grep interr /proc/interrupts
Also, can you try start FS with -nocal to see the difference?
I still using Ubuntu because it need to schedule down time to migrate to centos in our datacenter. So I really would like to see it works ok on Ubuntu.
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
These compariso is done into the same machine. I run SVN version, make the test, stop SVN version, run GIT version and make the test again.
It runs over a "Intel(R) Core(TM)2 Duo CPU E7500 @ 2.93GHz"
Regards
2010/8/6 Seven Du <dujinfang@gmail.com>
Were both of your kernel running on 1000HZ ?
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
I've got about 75% decrease in performarce. I am using same server and same config. DB was already in ramdisk as svn version so I don't find an explanation for it...
Regards
2010/8/5 Eliot Gable <egable+freeswitch@gmail.com<egable%2Bfreeswitch@gmail.com>
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the
new
GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
No one experiment this same issue?
I also noticed a high cpu consume on call hangup.
Regards
2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com> > > Hello! > > I was been some test using one FS tarball version downloaded some > weeks > ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some > problems > with > odbc connections managemens. Now I'm using a git version (FreeSWITCH > Version > 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to > solve > this > problem but I noticed it has a high cpu load comparing with svn > version. > While I could manage more than 200 calls with a 75% CPU load into a > dual > core server using svn version, now, 50 calls consume this 75% cpu. I > can see > same modules are loaded (except new hash module needed in git > version) > and > same scenario is used. I am not be able to find why now it uses more > cpu > than before. Any idea? > Each testing call is a simple bridge to an external sip provider. > > Regards
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE: http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
I cannot see updates/seconds using that command.
It should show number of interrupts every second
I can see following boot params for kernel at /boot/config-2.6.32-21-server:
CONFIG_NO_HZ=y CONFIG_HZ_100=y # CONFIG_HZ_250 is not set # CONFIG_HZ_300 is not set # CONFIG_HZ_1000 is not set CONFIG_HZ=100
That's it.
I have been reading about this task and people recomends using 1000 HZ instead 100 ¿what do you think?
Anthony recommends 1000 for best sound. I still use the default 250 on my box. Have you tried -nocal ? Again, I don't know if it's related with your problem.
Centos seems use 1000 by default.
Regards
2010/8/6 Seven Du <dujinfang@gmail.com>
The default Ubuntu kernel running on 100Hz/250Hz. And there might be some great changed in timing between your two versions. I don't remember exactly. And I don't know if it's the reason causing load.
I use the following command to see Hz, it should approximately increase 100 or 250 or 1000 each second.
watch grep interr /proc/interrupts
Also, can you try start FS with -nocal to see the difference?
I still using Ubuntu because it need to schedule down time to migrate to centos in our datacenter. So I really would like to see it works ok on Ubuntu.
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
These compariso is done into the same machine. I run SVN version, make the test, stop SVN version, run GIT version and make the test again.
It runs over a "Intel(R) Core(TM)2 Duo CPU E7500 @ 2.93GHz"
Regards
2010/8/6 Seven Du <dujinfang@gmail.com>
Were both of your kernel running on 1000HZ ?
2010/8/6 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>:
I've got about 75% decrease in performarce. I am using same server and same config. DB was already in ramdisk as svn version so I don't find an explanation for it...
Regards
2010/8/5 Eliot Gable <egable+freeswitch@gmail.com>
My most recent testing as of about two weeks ago showed a 25% performance drop between the old SVN version I was running and the new GIT version. Initially, it seemed closer to your reported performance drop, but after moving the database to a ramdisk, it went to 25%. I am unsure of whether my initial testing was using the db on a ramdisk, so the drop could be higher if my previous testing did not have the db on a ramdisk.
2010/8/3 Juan Antonio Ibañez Santorum <juanito1982@gmail.com>: > No one experiment this same issue? > > I also noticed a high cpu consume on call hangup. > > Regards > > 2010/8/2 Juan Antonio Ibañez Santorum <juanito1982@gmail.com> >> >> Hello! >> >> I was been some test using one FS tarball version downloaded some >> weeks >> ago (FreeSWITCH Version 1.0.6 (svn-exported)) but it has some >> problems >> with >> odbc connections managemens. Now I'm using a git version >> (FreeSWITCH >> Version >> 1.0.head (git-b485f25 2010-07-30 19-46-05 -0400)) that seems to >> solve >> this >> problem but I noticed it has a high cpu load comparing with svn >> version. >> While I could manage more than 200 calls with a 75% CPU load into >> a >> dual >> core server using svn version, now, 50 calls consume this 75% >> cpu. I >> can see >> same modules are loaded (except new hash module needed in git >> version) >> and >> same scenario is used. I am not be able to find why now it uses >> more >> cpu >> than before. Any idea? >> Each testing call is a simple bridge to an external sip provider. >> >> Regards > > > _______________________________________________ > FreeSWITCH-users mailing list > FreeSWITCH-users@lists.freeswitch.org > http://lists.freeswitch.org/mailman/listinfo/freeswitch-users > > > UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users > http://www.freeswitch.org > >
-- Eliot Gable
"We do not inherit the Earth from our ancestors: we borrow it from our children." ~David Brower
"I decided the words were too conservative for me. We're not borrowing from our children, we're stealing from them--and it's not even considered to be a crime." ~David Brower
"Esse oportet ut vivas, non vivere ut edas." (Thou shouldst eat to live; not live to eat.) ~Marcus Tullius Cicero
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users
UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
_______________________________________________ FreeSWITCH-users mailing list FreeSWITCH-users@lists.freeswitch.org http://lists.freeswitch.org/mailman/listinfo/freeswitch-users UNSUBSCRIBE:http://lists.freeswitch.org/mailman/options/freeswitch-users http://www.freeswitch.org
-- Blog: http://www.dujinfang.com Proj: http://www.freeswitch.org.cn
participants (7)
-
Brian West
-
Eliot Gable
-
Henry Huang
-
Juan Antonio Ibañez Santorum
-
Madovsky
-
Rupa Schomaker
-
Seven Du