Hi, This is just for people's information on my own experience in Zope memory leak. It may not reflect your personal case. But if it does, please post a note to the mailing list. When there is a problem, people have to face it. I really dislike the attitude of people who just shrug their shoulders. I have seen this all too many times. Everytime I go through a problem and posted in to the mailing list, there are always people that try to say nonsense and refuse to accept that there is a problem. Guess what? Weeks later more and more people suffer similar problems. By now I am very sure that a huge memory leak is coming from SQLSession. Remember about one or two months ago I mentioned about memory leak in Zope? There were so many die-hard Zope mongers that the very first thing they feel is their pride being hurt, and instead of looking at possible sources of the problem, their attitude was: it can't be, you must be doing something wrong, Zope has no memory leak. Now, does that attitude help to find the source of the problem???! Perl mongers are bad enough, and now we have Zope mongers. The thing is, no one would complain about memory leak if there is no performance impact. When someone complains about memory leak, it's because the performance is getting to be unbearable. After using SQLSession with two different databases (Gadfly and Sybase), and after the problem persisted, and after using manage_debug in Control Panel, I have decided that enough is enough. I implemented my own RAM-based Session Management. And it's been about a month now. No more memory leak. Zope memory usage stayed nice and low. Performance improved substantially to the amazement of everyone. No more need to re-start Zope to bring the memory down. There are hundreds of people out there using SQLSession. I know for fact that someone using SQLSession has actually written a Unix cron job to restart Zope every night in order to keep the memory usage down. (Of course, the person does not know the memory leak is coming from SQLSession.)
From the manage_debug screen, it seems that the leak has something to do with the SQLSession inheriting from Implicit class. I don't remember much anymore (since I got rid of SQLSession already a while ago.) I just remember the keyword "Implicit" showing up in the objects in the manage_debug screen.
All I know is that after I stop using SQLSession (mainly with Gadfly, which was a temporary solution anyway), my problems stopped. regards, Hung Jung ________________________________________________________________________ Get Your Private, Free E-mail from MSN Hotmail at http://www.hotmail.com
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1 In article <20000526191004.49713.qmail@hotmail.com>, Hung Jung Lu <hungjunglu@hotmail.com> writes
After using SQLSession with two different databases (Gadfly and Sybase), and after the problem persisted, and after using manage_debug in Control Panel, I have decided that enough is enough. I implemented my own RAM-based Session Management.
I recall you mentioning the memory leak with Gadfly, but do not recall you mentioning it again until now. I've also noticed a memory leak under WinNT and I use SQLSession with Interbase. But I assumed it was because I was also using Gadfly for a navigational tool. And since I have to reboot once every few weeks, I wasn't too concerned about it. Will you be releasing your Session management as a product? - -- Regards, Graham Chiu gchiu<at>compkarori.co.nz http://www.compkarori.co.nz/index.php Powered by Interbase and Zope -----BEGIN PGP SIGNATURE----- Version: PGPsdk version 1.7.1 iQA/AwUBOS4zALTRdIWzaLpMEQLuqgCcDJimO+vxzZIOyDl1S/1Fmrut/cQAn2J5 aGGcAP28XWP+ZRFbL387CiLo =eSu9 -----END PGP SIGNATURE-----
Hung Jung Lu wrote:
Hi,
This is just for people's information on my own experience in Zope memory leak. It may not reflect your personal case. But if it does, please post a note to the mailing list.
When there is a problem, people have to face it. I really dislike the attitude of people who just shrug their shoulders. I have seen this all too many times. Everytime I go through a problem and posted in to the mailing list, there are always people that try to say nonsense and refuse to accept that there is a problem. Guess what? Weeks later more and more people suffer similar problems.
Hung, if you recall, there were lots of people who when someone mentioned a memory leak, looked at their process and misinterpreted the threads' RAM reports. To be honest, it is exceedingly difficult to track down memory leaks when you get an abundance of misinformation (I know, not intentional, but is still makes it hard). Memory leaks are of prime concern to me, as I have mentioned. I apologize if I came of as being someone who thought there was _no_ memory leak, I just needed accurate info to help. We were getting an increasing number of people who suddenly thought they had memory leaks because of the way Linux reports threads. Information like you have just provided (that you tracked it down to SQLSession) is very useful, thanks for providing it. Though not the maintainer, I will look into it.
There are hundreds of people out there using SQLSession. I know for fact that someone using SQLSession has actually written a Unix cron job to restart Zope every night in order to keep the memory usage down. (Of course, the person does not know the memory leak is coming from SQLSession.)
Why not help them out and let them know? ;-) I believe you can switch from SQLSession to FSSEssion with little, if any DTML changes.
From the manage_debug screen, it seems that the leak has something to do with the SQLSession inheriting from Implicit class. I don't remember much anymore (since I got rid of SQLSession already a while ago.) I just remember the keyword "Implicit" showing up in the objects in the manage_debug screen.
All I know is that after I stop using SQLSession (mainly with Gadfly, which was a temporary solution anyway), my problems stopped.
To be fair, Hung, SQLSession is not Zope, it is a third party product. I would also send the information to the maintainer of SQLSession, so he/she can be aware of it.
On Fri, May 26, 2000 at 12:10:04PM -0700, Hung Jung Lu wrote:
This is just for people's information on my own experience in Zope memory leak. It may not reflect your personal case. But if it does, please post a note to the mailing list.
When there is a problem, people have to face it. I really dislike the attitude of people who just shrug their shoulders. I have seen this all too many times. Everytime I go through a problem and posted in to the mailing list, there are always people that try to say nonsense and refuse to accept that there is a problem. Guess what? Weeks later more and more people suffer similar problems.
Could you point us to archived posts that support this claim? I think that you missed the efforts on Zope-dev and a Zope.org Wiki on the subject where some very dedicated people were/are trying to find out if 1: what reported cases of alleged memory leaks are in fact genuine, and 2: what is/are the cause(s) of any memory leaks that _do_ exist. Could you please report some more detail about your setup and testing methods to that group of people? It might be of some help to them. It would be more constructive. -- Martijn Pieters | Software Engineer mailto:mj@digicool.com | Digital Creations http://www.digicool.com/ | Creators of Zope http://www.zope.org/ | The Open Source Web Application Server ---------------------------------------------
participants (4)
-
Bill Anderson -
Graham Chiu -
Hung Jung Lu -
Martijn Pieters