- Pharo-dev - lists.pharo.org

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Igor Stasenko

On 15 July 2011 12:39, Guido Stepken <gstepken(a)googlemail.com> wrote: > I am doing server tuning, building up fast server systems for a very long time now. > > Yesterday i installed 1.2.1, expecting to see a far more matured Pharo Smalltalk, stable, pretty nice new features. > > I opened up the Memory Grapher, hmmm.... works fine, watching memory pumping. A bit strange, seeing a Pharo consuming and freeing megabytes of memory space without Â any load. But ok! > > Then i opened up TestSuites, marked some tests, let them run. Ups! Desktop hanging. > > I expected to have with Pharo a well done, stable multitasking environment, but Pharo hangs everywhere, is leaking lots of memory, is keeping up connections, wasting resources unneccessarily, having problems with internal race conditions on semaphores, mutexes, feels same like "Big Kernel Lock", but even without MultiProcessing. > > The Pharo Team IMHO is very much missing important knowledge about modern Operating System Design, what Pharo Smalltalk OS is in fact, without being able to fulfill the minimum standards i expect from a usable Smalltalk OS. > > As system operator or enterprise i cannot offer such a system, that has to be restarted twenty times a day, due to hangs, memory leaks, sucking connection handles, ..., whatever! > > tnx for your understanding! > > regards, Guido Stepken > >From an outsider point of view, your message is ok: "i took a glance on system and found it inappropriate for my tasks, so i will not use it". Fine. Now, from an insider point of view, your message is worthless, because it doesn't contains any concrete proposals to make things better (to your thinking). We would like to improve our system to make it good enough for people like you. But we need help. So, every time you saying 'i don't like X', you should also argument that and propose better solution. If you not doing that, but instead using vague statements, you just wasting own and our time. Lets be concrete and constructive. Pharo is far from perfect. And everyone understands that. You made no discovery. Now, ask yourself how your message helps to improve that? -- Best regards, Igor Stasenko AKA sig.

July 15, 2011

[Pharo-project] A request or suggestion.

by Friedrich Dominicus

I'm still not realyl understanding the pharo Morph stuff. So I just want to ask. Would it be possible/feasable to group windows. E.g I'd have a group of Workspaces a group of Browsers etc. Currently if the windows just pop open it's hard to find a workspace one may be interested. If there would be a group this should be much easier. Regards Friedrich

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Sven Van Caekenberghe

On 15 Jul 2011, at 13:50, Janko MivÅ¡ek wrote: > Before shooting the messenger of bad news let we rather acknowledge that > there are problems in Squeak/Pharo on the socket/semaphore/process > field. I also admit, I somewhat lost nervers on Pharo as Guido did and > I'd like to post the same post. But of course bringing bad messages must > always be done diplomatically of you won't like to be shoot :) Janko, you have an excellent Smalltalk reputation and your comments, feedback and contribution are positive. You walk the talk. We take you seriously. > So, known problems are: socket limit over about 200, semaphores (i > remember discussion back, is this solved?) protecting critical sections > doesn't work well, processes? After our last discussion, I added ! "Extend the external object table to allow for more sockets" Smalltalk vmParameterAt: 49 put: 8192. ! to my server startup scripts. This should be good for at least around 2000 sockets simulataneously, if I understand things correctly. (Your OS has to allow that many for a single process as well ;-) Is there a bug report/issue and/or testcase for the critical section problem ? Regards, Sven

July 15, 2011

Re: [Pharo-project] Fixed up AutomaticMethodCategorizer

by StÃ©phane Ducasse

BTW I would love to recategorize all the methods in the image :) Stef On Jul 15, 2011, at 1:48 PM, Mariano Martinez Peck wrote: > Thanks Danie. I don't have time to take a look right now, but maybe someone can. Anyway, you may know AutomaticMethodCategorizer much better than us. So you can just go ahead and commit and update ConfifugrationOfAutomaticMethodCategorizer > > On Wed, Jul 13, 2011 at 5:27 PM, Danie Roux <danie(a)danieroux.com> wrote: > Hi Mariano, > > We need the capability to set custom rules on > AutomaticMethodCategorizer. I've added it back in > http://www.squeaksource.com/autoMethodCat/AutomaticMethodCategorizer-DanieR… > > If you will please test it and update the Configuration, I will appreciate it. > > -- > Danie Roux *shuffle* Adore Unix - http://danieroux.com > Email / Google Talk / Jabber - danie(a)danieroux.com > > > > -- > Mariano > http://marianopeck.wordpress.com >

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Norbert Hartl

Am 15.07.2011 um 12:39 schrieb Guido Stepken: > As system operator or enterprise i cannot offer such a system, that has to be restarted twenty times a day, due to hangs, memory leaks, sucking connection handles, ..., whatever! Sorry, but that made me laugh. And I'm really asking myself if you ever worked in big industries. If you think like this you shouldn't ever dare to, you will asking for nightmares. couldn't resist, Norbert

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Janko MivÅ¡ek

Hi all, Before shooting the messenger of bad news let we rather acknowledge that there are problems in Squeak/Pharo on the socket/semaphore/process field. I also admit, I somewhat lost nervers on Pharo as Guido did and I'd like to post the same post. But of course bringing bad messages must always be done diplomatically of you won't like to be shoot :) So, known problems are: socket limit over about 200, semaphores (i remember discussion back, is this solved?) protecting critical sections doesn't work well, processes? So Guido is right, don't shut them, let his anger rather energitize us to finally solve them! Janko S, StÃ©phane Ducasse piÅ¡e: > Sven from beta9 is using Pharo as a server and it works well from what he told us during the last sprint. > Netstyle is also using pharo for Seaside. > > Now there is certainly place for improvements and pharo is improving every day (but because there are cool pharoers improving it). > > I noticed in the past that you always have this same aggressive tone. > And even worse, we value critics and non politically correct discussions but not systematic bashing. > What I dislike is the tone of your email. So do not continue like that if you want to have any support for people. > else you will be plonkified :) > >> I am doing server tuning, building up fast server systems for a very long time now. >> >> Yesterday i installed 1.2.1, expecting to see a far more matured Pharo Smalltalk, stable, pretty nice new features. >> >> I opened up the Memory Grapher, hmmm.... works fine, watching memory pumping. A bit strange, seeing a Pharo consuming and freeing megabytes of memory space without any load. But ok! >> >> Then i opened up TestSuites, marked some tests, let them run. Ups! Desktop hanging. >> >> I expected to have with Pharo a well done, stable multitasking environment, but Pharo hangs everywhere, is leaking lots of memory, is keeping up connections, wasting resources unneccessarily, having problems with internal race conditions on semaphores, mutexes, feels same like "Big Kernel Lock", but even without MultiProcessing. > > Only that. Not bad. We are soo dumb that it could be so worse. > >> The Pharo Team IMHO is very much missing important knowledge about modern Operating System Design, what Pharo Smalltalk OS is in fact, without being able to fulfill the minimum standards i expect from a usable Smalltalk OS. >> >> As system operator or enterprise i cannot offer such a system, that has to be restarted twenty times a day, due to hangs, memory leaks, sucking connection handles, ..., whatever! >> >> tnx for your understanding! > > My understanding is that you do not know how to communicate. > > Now my advice is to try 1.3 because lot of improvements were applied and for example Zinc is a good starting part. > >> >> regards, Guido Stepken > > > -- Janko MivÅ¡ek Aida/Web Smalltalk Web Application Server http://www.aidaweb.si

July 15, 2011

Re: [Pharo-project] Fixed up AutomaticMethodCategorizer

by Mariano Martinez Peck

Thanks Danie. I don't have time to take a look right now, but maybe someone can. Anyway, you may know AutomaticMethodCategorizer much better than us. So you can just go ahead and commit and update ConfifugrationOfAutomaticMethodCategorizer On Wed, Jul 13, 2011 at 5:27 PM, Danie Roux <danie(a)danieroux.com> wrote: > Hi Mariano, > > We need the capability to set custom rules on > AutomaticMethodCategorizer. I've added it back in > > http://www.squeaksource.com/autoMethodCat/AutomaticMethodCategorizer-DanieR… > > If you will please test it and update the Configuration, I will appreciate > it. > > -- > Danie Roux *shuffle* Adore Unix - http://danieroux.com > Email / Google Talk / Jabber - danie(a)danieroux.com > -- Mariano http://marianopeck.wordpress.com

July 15, 2011

Re: [Pharo-project] Zinc time outs too fast

by Pavel Krivanek

Hi Sven, On Fri, Jul 15, 2011 at 1:01 PM, Sven Van Caekenberghe <sven(a)beta9.be> wrote: > Pavel, Igor, > > On 14 Jul 2011, at 10:06, Sven Van Caekenberghe wrote: > >> There could still be some deep problem somewhere (you say the default timeout of 30 shows up in stack), but as far as I can see this works. >> >> I have said this before, ConnectionTimedOut should in 99% of the case be interpreted as ResourceUnavailable, 30 seconds is an eternity for all but a small number of data or processing heavy requests. > > Does it work now ? > Yes, we added longer timeouts for download of package contents and it is working now well. Thank you. -- Pavel > Sven > > > >

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by StÃ©phane Ducasse

Sven from beta9 is using Pharo as a server and it works well from what he told us during the last sprint. Netstyle is also using pharo for Seaside. Now there is certainly place for improvements and pharo is improving every day (but because there are cool pharoers improving it). I noticed in the past that you always have this same aggressive tone. And even worse, we value critics and non politically correct discussions but not systematic bashing. What I dislike is the tone of your email. So do not continue like that if you want to have any support for people. else you will be plonkified :) > I am doing server tuning, building up fast server systems for a very long time now. > > Yesterday i installed 1.2.1, expecting to see a far more matured Pharo Smalltalk, stable, pretty nice new features. > > I opened up the Memory Grapher, hmmm.... works fine, watching memory pumping. A bit strange, seeing a Pharo consuming and freeing megabytes of memory space without any load. But ok! > > Then i opened up TestSuites, marked some tests, let them run. Ups! Desktop hanging. > > I expected to have with Pharo a well done, stable multitasking environment, but Pharo hangs everywhere, is leaking lots of memory, is keeping up connections, wasting resources unneccessarily, having problems with internal race conditions on semaphores, mutexes, feels same like "Big Kernel Lock", but even without MultiProcessing. Only that. Not bad. We are soo dumb that it could be so worse. > The Pharo Team IMHO is very much missing important knowledge about modern Operating System Design, what Pharo Smalltalk OS is in fact, without being able to fulfill the minimum standards i expect from a usable Smalltalk OS. > > As system operator or enterprise i cannot offer such a system, that has to be restarted twenty times a day, due to hangs, memory leaks, sucking connection handles, ..., whatever! > > tnx for your understanding! My understanding is that you do not know how to communicate. Now my advice is to try 1.3 because lot of improvements were applied and for example Zinc is a good starting part. > > regards, Guido Stepken

July 15, 2011

Re: [Pharo-project] Hanging connects, exhausted resources, memory leaks, bocked everything ... Unusable!

by Luc Fabresse

Hi Guido, I am always interested when some companies try to really use Pharo in production. Some already does. If there are issues we would be glad to improve the system and correct them. But, I never experienced all you stated certainly because we have different environments and uses. So please, give us real feedback: - which VM did you used - which OS - which tests hang your OS - ... Regards, #Luc 2011/7/15 Carlo <snoobabk(a)yahoo.ie> > Hi Guido > > How did you come to all your conclusions of Pharo if you simply installed > and opened up Pharo 1.2.1 and only ran the tests? > Memory leaks, connections, unusable etc??? > > If you have any explicit and constructive feedback then the Pharo community > can at least use that to improve it.... > > Cheers > Carlo > > On 15 Jul 2011, at 12:39 PM, Guido Stepken wrote: > > I am doing server tuning, building up fast server systems for a very long > time now. > > Yesterday i installed 1.2.1, expecting to see a far more matured Pharo > Smalltalk, stable, pretty nice new features. > > I opened up the Memory Grapher, hmmm.... works fine, watching memory > pumping. A bit strange, seeing a Pharo consuming and freeing megabytes of > memory space without any load. But ok! > > Then i opened up TestSuites, marked some tests, let them run. Ups! Desktop > hanging. > > I expected to have with Pharo a well done, stable multitasking environment, > but Pharo hangs everywhere, is leaking lots of memory, is keeping up > connections, wasting resources unneccessarily, having problems with internal > race conditions on semaphores, mutexes, feels same like "Big Kernel Lock", > but even without MultiProcessing. > > The Pharo Team IMHO is very much missing important knowledge about modern > Operating System Design, what Pharo Smalltalk OS is in fact, without being > able to fulfill the minimum standards i expect from a usable Smalltalk OS. > > As system operator or enterprise i cannot offer such a system, that has to > be restarted twenty times a day, due to hangs, memory leaks, sucking > connection handles, ..., whatever! > > tnx for your understanding! > > regards, Guido Stepken > > >

July 15, 2011