Well, we have been having problems with
applications appearing to hang every 6 days or so (sound
familiar?).
I have identified a bug in the Delay
implementation, rather obscure, but highlighted by our applications that make
HEAVY use of processes/(tight, a few tens of milliseconds)delays/semaphores
etc.
Naturally due to an edge case on millisecond clock
rollover...
Affects all Squeak/Pharo versions, as far as I can
tell. (It is separate from the OSProcess issues I have posted about a few weeks
ago).
I've developed a potential fix, though I'll have to
let you all know whether (hopefully) successful in a week or two... fingers
crossed!
I hope it is fixable, last hurdle for business
applications, from our point of view.
I'll post details when fix confirmed or failed
then.
Regards, Gary |
thanks gary.
Yes adrian got a problem on his server after two weeks of load. I do not know how we can improve them. Stef On Apr 6, 2011, at 6:02 PM, Gary Chambers wrote: > Well, we have been having problems with applications appearing to hang every 6 days or so (sound familiar?). > > I have identified a bug in the Delay implementation, rather obscure, but highlighted by our applications that make HEAVY use of processes/(tight, a few tens of milliseconds)delays/semaphores etc. > Naturally due to an edge case on millisecond clock rollover... > > Affects all Squeak/Pharo versions, as far as I can tell. (It is separate from the OSProcess issues I have posted about a few weeks ago). > > I've developed a potential fix, though I'll have to let you all know whether (hopefully) successful in a week or two... fingers crossed! > > I hope it is fixable, last hurdle for business applications, from our point of view. > > I'll post details when fix confirmed or failed then. > > Regards, Gary |
Bear with me, hopefully my fix will work (yay for all then, especially
Seaside peeps!). Regards, Gary ----- Original Message ----- From: "Stéphane Ducasse" <[hidden email]> To: <[hidden email]> Sent: Wednesday, April 06, 2011 5:24 PM Subject: Re: [Pharo-project] Delay bug thanks gary. Yes adrian got a problem on his server after two weeks of load. I do not know how we can improve them. Stef On Apr 6, 2011, at 6:02 PM, Gary Chambers wrote: > Well, we have been having problems with applications appearing to hang > every 6 days or so (sound familiar?). > > I have identified a bug in the Delay implementation, rather obscure, but > highlighted by our applications that make HEAVY use of processes/(tight, a > few tens of milliseconds)delays/semaphores etc. > Naturally due to an edge case on millisecond clock rollover... > > Affects all Squeak/Pharo versions, as far as I can tell. (It is separate > from the OSProcess issues I have posted about a few weeks ago). > > I've developed a potential fix, though I'll have to let you all know > whether (hopefully) successful in a week or two... fingers crossed! > > I hope it is fixable, last hurdle for business applications, from our > point of view. > > I'll post details when fix confirmed or failed then. > > Regards, Gary |
In reply to this post by Stéphane Ducasse
We managed to resolve our initial problem with the process scheduling [1]. All Cmsbox customers have been migrated to Pharo and it works very reliably now.
Gary, maybe you could open a ticket already now and post what you know at this time to get early feedback. [1] http://code.google.com/p/pharo/issues/detail?id=3498 Cheers Adrian On Apr 6, 2011, at 18:24 , Stéphane Ducasse wrote: > thanks gary. > Yes adrian got a problem on his server after two weeks of load. I do not know how we can improve them. > > Stef > > On Apr 6, 2011, at 6:02 PM, Gary Chambers wrote: > >> Well, we have been having problems with applications appearing to hang every 6 days or so (sound familiar?). >> >> I have identified a bug in the Delay implementation, rather obscure, but highlighted by our applications that make HEAVY use of processes/(tight, a few tens of milliseconds)delays/semaphores etc. >> Naturally due to an edge case on millisecond clock rollover... >> >> Affects all Squeak/Pharo versions, as far as I can tell. (It is separate from the OSProcess issues I have posted about a few weeks ago). >> >> I've developed a potential fix, though I'll have to let you all know whether (hopefully) successful in a week or two... fingers crossed! >> >> I hope it is fixable, last hurdle for business applications, from our point of view. >> >> I'll post details when fix confirmed or failed then. >> >> Regards, Gary > > |
Issue http://code.google.com/p/pharo/issues/detail?id=3986
Regards, Gary ----- Original Message ----- From: "Adrian Lienhard" <[hidden email]> To: <[hidden email]> Sent: Wednesday, April 06, 2011 5:41 PM Subject: Re: [Pharo-project] Delay bug We managed to resolve our initial problem with the process scheduling [1]. All Cmsbox customers have been migrated to Pharo and it works very reliably now. Gary, maybe you could open a ticket already now and post what you know at this time to get early feedback. [1] http://code.google.com/p/pharo/issues/detail?id=3498 Cheers Adrian On Apr 6, 2011, at 18:24 , Stéphane Ducasse wrote: > thanks gary. > Yes adrian got a problem on his server after two weeks of load. I do not > know how we can improve them. > > Stef > > On Apr 6, 2011, at 6:02 PM, Gary Chambers wrote: > >> Well, we have been having problems with applications appearing to hang >> every 6 days or so (sound familiar?). >> >> I have identified a bug in the Delay implementation, rather obscure, but >> highlighted by our applications that make HEAVY use of processes/(tight, >> a few tens of milliseconds)delays/semaphores etc. >> Naturally due to an edge case on millisecond clock rollover... >> >> Affects all Squeak/Pharo versions, as far as I can tell. (It is separate >> from the OSProcess issues I have posted about a few weeks ago). >> >> I've developed a potential fix, though I'll have to let you all know >> whether (hopefully) successful in a week or two... fingers crossed! >> >> I hope it is fixable, last hurdle for business applications, from our >> point of view. >> >> I'll post details when fix confirmed or failed then. >> >> Regards, Gary > > |
Free forum by Nabble | Edit this page |