You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@syncope.apache.org by Adrian Gonzalez <ad...@yahoo.fr> on 2016/08/29 14:04:50 UTC
Syncope startup performance improvements
Hello,
I'm using syncope 2.0.0-M4.Standalone startup takes 135s on my laptop.
I would like to improve syncope's startup.
I've made some experimentations :
1. Restricting scanning on basePackages I've changed ClassPathScanImplementationLookup to scan only some specific packages. This results in a 32s startup.
https://github.com/apache/syncope/pull/36
2. Adding startStopThreads="5" in server.xml This results in a 26s startup.
3. I've also tried to set <JarScanner scanClassPath="false"/> in context.xml, but no improvement.
4. I've further modified ClassPathScanImplementationLookup to scan a fixed list of classes. As is, this doesn't work. I would have a 20.7s startup (cumulated with change 2) https://github.com/gonzalad/syncope/tree/2.0.0-M2-startup-perf2
Could it be possible to incorporate first PR in syncope code (pull/36) ? Sorry, I made PR on 2.0.0-M4, but could make 2.0.0-SNASPHOT work
Are you interested in improvement #4 ? (in this case, I can spend a bit more time on it)
Thanks,Adrian
Re: Syncope startup performance improvements
Posted by Francesco Chicchiriccò <il...@apache.org>.
Not to me, please follow the linked instructions, thanks.
Il 30 agosto 2016 10:18:24 CEST, Adrian Gonzalez <ad...@yahoo.fr> ha scritto:
>Hi Francesco,
>I've signed the ICLA (sent you in pm), and created
>issue�https://issues.apache.org/jira/browse/SYNCOPE-926
>Thanks for considering the fix !Adrian
>
> De�: Francesco Chicchiricc� <il...@apache.org>
> ��: user@syncope.apache.org
> Envoy� le : Lundi 29 ao�t 2016 18h47
> Objet�: Re: Syncope startup performance improvements
>
>Hi,
>your improvement looks interesting; would you mind:
>
>1. sending an ICLA, as explained in the last paragraph from
>
>http://syncope.apache.org/contributing.html
>
>2. opening an improvement on JIRA?
>
>I'll take a deeper look once back at work, this Thursday.
>
>Thanks for your contribution.
>Regards.
>
>On 29 aug 2016 16:04:50 CEST, Adrian Gonzalez <ad...@yahoo.fr>
>wrote:
>>Hello,
>>I'm using syncope 2.0.0-M4.Standalone startup takes 135s on my laptop.
>>I would like to improve syncope's startup.
>>I've made some experimentations :
>>1. Restricting scanning on basePackages��� I've changed
>>ClassPathScanImplementationLookup to scan only some specific
>>packages.��� This results in a 32s startup.
>>��� https://github.com/apache/syncope/pull/36
>>
>>2. Adding startStopThreads="5" in server.xml��� This results in a 26s
>>startup.
>>3. I've also tried to set <JarScanner scanClassPath="false"/> in
>>context.xml, but no improvement.
>>4. I've further modified ClassPathScanImplementationLookup to scan a
>>fixed list of classes.�� As is, this doesn't work.�� I would have a
>>20.7s startup (cumulated with change 2)��
>>https://github.com/gonzalad/syncope/tree/2.0.0-M2-startup-perf2
>>
>>
>>Could it be possible to incorporate first PR in syncope code (pull/36)
>>? Sorry, I made PR on 2.0.0-M4, but could make 2.0.0-SNASPHOT work
>>Are you interested in improvement #4 ? (in this case, I can spend a
>bit
>>more time on it)
>>Thanks,Adrian
--
Francesco Chicchiricc�
Tirasa - Open Source Excellence
http://www.tirasa.net/
Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
CXF Committer, OpenJPA Committer
http://home.apache.org/~ilgrosso/
Re: Syncope startup performance improvements
Posted by Adrian Gonzalez <ad...@yahoo.fr>.
Hi Francesco,
I've signed the ICLA (sent you in pm), and created issue https://issues.apache.org/jira/browse/SYNCOPE-926
Thanks for considering the fix !Adrian
De : Francesco Chicchiriccò <il...@apache.org>
À : user@syncope.apache.org
Envoyé le : Lundi 29 août 2016 18h47
Objet : Re: Syncope startup performance improvements
Hi,
your improvement looks interesting; would you mind:
1. sending an ICLA, as explained in the last paragraph from
http://syncope.apache.org/contributing.html
2. opening an improvement on JIRA?
I'll take a deeper look once back at work, this Thursday.
Thanks for your contribution.
Regards.
On 29 aug 2016 16:04:50 CEST, Adrian Gonzalez <ad...@yahoo.fr> wrote:
>Hello,
>I'm using syncope 2.0.0-M4.Standalone startup takes 135s on my laptop.
>I would like to improve syncope's startup.
>I've made some experimentations :
>1. Restricting scanning on basePackages I've changed
>ClassPathScanImplementationLookup to scan only some specific
>packages. This results in a 32s startup.
> https://github.com/apache/syncope/pull/36
>
>2. Adding startStopThreads="5" in server.xml This results in a 26s
>startup.
>3. I've also tried to set <JarScanner scanClassPath="false"/> in
>context.xml, but no improvement.
>4. I've further modified ClassPathScanImplementationLookup to scan a
>fixed list of classes. As is, this doesn't work. I would have a
>20.7s startup (cumulated with change 2)
>https://github.com/gonzalad/syncope/tree/2.0.0-M2-startup-perf2
>
>
>Could it be possible to incorporate first PR in syncope code (pull/36)
>? Sorry, I made PR on 2.0.0-M4, but could make 2.0.0-SNASPHOT work
>Are you interested in improvement #4 ? (in this case, I can spend a bit
>more time on it)
>Thanks,Adrian
--
Francesco Chicchiriccò
Tirasa - Open Source Excellence
http://www.tirasa.net/
Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
CXF Committer, OpenJPA Committer
http://home.apache.org/~ilgrosso/
Re: Syncope startup performance improvements
Posted by Francesco Chicchiriccò <il...@apache.org>.
Hi Adrian,
it seems that your fix was effective, then: thanks for reporting!
On 07/09/2016 09:19, Adrian Gonzalez wrote:
> Hi,
>
> Just a little feedback on 2.0.0-M5 and SYNCOPE-926 :
> * we've upgraded to syncope-2.0.0-SNAPSHOT (past M5) it works fine,
> so far so good ;)
> * with the improvements made in 2.0.0-M5 and in SYNCOPE-926, we
> changed our startup speed from 215s to 24s.
> Note that our tomcat server contains both syncope and cxf, and I
> also added scanClassPath="false" and startStopThreads="5" to our
> tomcat configuration.
>
> Thanks once more !
> Adrian
>
> ------------------------------------------------------------------------
> *De :* Francesco Chicchiricc� <il...@apache.org>
> *� :* user@syncope.apache.org
> *Envoy� le :* Lundi 29 ao�t 2016 18h47
> *Objet :* Re: Syncope startup performance improvements
>
> Hi,
> your improvement looks interesting; would you mind:
>
> 1. sending an ICLA, as explained in the last paragraph from
>
> http://syncope.apache.org/contributing.html
>
> 2. opening an improvement on JIRA?
>
> I'll take a deeper look once back at work, this Thursday.
>
> Thanks for your contribution.
> Regards.
>
> On 29 aug 2016 16:04:50 CEST, Adrian Gonzalez <adr_gonzalez@yahoo.fr
> <ma...@yahoo.fr>> wrote:
> >Hello,
> >I'm using syncope 2.0.0-M4.Standalone startup takes 135s on my laptop.
> >I would like to improve syncope's startup.
> >I've made some experimentations :
> >1. Restricting scanning on basePackages I've changed
> >ClassPathScanImplementationLookup to scan only some specific
> >packages. This results in a 32s startup.
> > https://github.com/apache/syncope/pull/36
> >
> >2. Adding startStopThreads="5" in server.xml This results in a 26s
> >startup.
> >3. I've also tried to set <JarScanner scanClassPath="false"/> in
> >context.xml, but no improvement.
> >4. I've further modified ClassPathScanImplementationLookup to scan a
> >fixed list of classes. As is, this doesn't work. I would have a
> >20.7s startup (cumulated with change 2)
> >https://github.com/gonzalad/syncope/tree/2.0.0-M2-startup-perf2
> >
> >
> >Could it be possible to incorporate first PR in syncope code (pull/36)
> >? Sorry, I made PR on 2.0.0-M4, but could make 2.0.0-SNASPHOT work
> >Are you interested in improvement #4 ? (in this case, I can spend a bit
> >more time on it)
> >Thanks,Adrian
--
Francesco Chicchiricc�
Tirasa - Open Source Excellence
http://www.tirasa.net/
Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
CXF Committer, OpenJPA Committer, PonyMail PPMC
http://home.apache.org/~ilgrosso/
Re: Syncope startup performance improvements
Posted by Adrian Gonzalez <ad...@yahoo.fr>.
Hi,
Just a little feedback on 2.0.0-M5 and SYNCOPE-926 :
* we've upgraded to syncope-2.0.0-SNAPSHOT (past M5) it works fine, so far so good ;)
* with the improvements made in 2.0.0-M5 and in SYNCOPE-926, we changed our startup speed from 215s to 24s.
Note that our tomcat server contains both syncope and cxf, and I also added scanClassPath="false" and startStopThreads="5" to our tomcat configuration.
Thanks once more !Adrian
De : Francesco Chicchiriccò <il...@apache.org>
À : user@syncope.apache.org
Envoyé le : Lundi 29 août 2016 18h47
Objet : Re: Syncope startup performance improvements
Hi,
your improvement looks interesting; would you mind:
1. sending an ICLA, as explained in the last paragraph from
http://syncope.apache.org/contributing.html
2. opening an improvement on JIRA?
I'll take a deeper look once back at work, this Thursday.
Thanks for your contribution.
Regards.
On 29 aug 2016 16:04:50 CEST, Adrian Gonzalez <ad...@yahoo.fr> wrote:
>Hello,
>I'm using syncope 2.0.0-M4.Standalone startup takes 135s on my laptop.
>I would like to improve syncope's startup.
>I've made some experimentations :
>1. Restricting scanning on basePackages I've changed
>ClassPathScanImplementationLookup to scan only some specific
>packages. This results in a 32s startup.
> https://github.com/apache/syncope/pull/36
>
>2. Adding startStopThreads="5" in server.xml This results in a 26s
>startup.
>3. I've also tried to set <JarScanner scanClassPath="false"/> in
>context.xml, but no improvement.
>4. I've further modified ClassPathScanImplementationLookup to scan a
>fixed list of classes. As is, this doesn't work. I would have a
>20.7s startup (cumulated with change 2)
>https://github.com/gonzalad/syncope/tree/2.0.0-M2-startup-perf2
>
>
>Could it be possible to incorporate first PR in syncope code (pull/36)
>? Sorry, I made PR on 2.0.0-M4, but could make 2.0.0-SNASPHOT work
>Are you interested in improvement #4 ? (in this case, I can spend a bit
>more time on it)
>Thanks,Adrian
--
Francesco Chicchiriccò
Tirasa - Open Source Excellence
http://www.tirasa.net/
Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
CXF Committer, OpenJPA Committer
http://home.apache.org/~ilgrosso/
Re: Syncope startup performance improvements
Posted by Francesco Chicchiriccò <il...@apache.org>.
Hi,
your improvement looks interesting; would you mind:
1. sending an ICLA, as explained in the last paragraph from
http://syncope.apache.org/contributing.html
2. opening an improvement on JIRA?
I'll take a deeper look once back at work, this Thursday.
Thanks for your contribution.
Regards.
On 29 aug 2016 16:04:50 CEST, Adrian Gonzalez <ad...@yahoo.fr> wrote:
>Hello,
>I'm using syncope 2.0.0-M4.Standalone startup takes 135s on my laptop.
>I would like to improve syncope's startup.
>I've made some experimentations :
>1. Restricting scanning on basePackages I've changed
>ClassPathScanImplementationLookup to scan only some specific
>packages. This results in a 32s startup.
> https://github.com/apache/syncope/pull/36
>
>2. Adding startStopThreads="5" in server.xml This results in a 26s
>startup.
>3. I've also tried to set <JarScanner scanClassPath="false"/> in
>context.xml, but no improvement.
>4. I've further modified ClassPathScanImplementationLookup to scan a
>fixed list of classes. As is, this doesn't work. I would have a
>20.7s startup (cumulated with change 2)
>https://github.com/gonzalad/syncope/tree/2.0.0-M2-startup-perf2
>
>
>Could it be possible to incorporate first PR in syncope code (pull/36)
>? Sorry, I made PR on 2.0.0-M4, but could make 2.0.0-SNASPHOT work
>Are you interested in improvement #4 ? (in this case, I can spend a bit
>more time on it)
>Thanks,Adrian
--
Francesco Chicchiriccò
Tirasa - Open Source Excellence
http://www.tirasa.net/
Involved at The Apache Software Foundation:
member, Syncope PMC chair, Cocoon PMC, Olingo PMC,
CXF Committer, OpenJPA Committer
http://home.apache.org/~ilgrosso/