Re: WebCatalog Eating 200% of the CPU
This WebDNA talk-list message is from 2002
It keeps the original formatting.
numero = 44949
interpreted = N
texte = Are you talking about exclusivelock or are there other 'tricks' as well?On 11/5/02 11:11 AM, Scott Anderson mashed the following keys :> What do you mean by 'clean'? A template may contain valid WebDNA syntax,> but could still take a long time to process. I would consider such a> template 'not' a very good candidate for a generally accessible WebDNA> template.> > Your argument assumes that you (or a hacker) have knowledge of a particular> template (or were able to create a template) on the server that can tie up> the engine enough to overwhelm it by sending several http requests for that> template. If no such template existed, then there is not a risk. Again,> this boils down to how carefully and efficiently you have programmed your> templates.> > If it is necessary to create a WebDNA template that will take a large amount> of time to complete (often some admin process), then you should put the> necessary WebDNA code in place to synchronize access to that template to> ensure that only one instance of that template will run at any given time.> Templates that you serve to the general public should be 'lean and mean' and> not 'over loaded' with to much code.> >> why have the preference?> A high performance server can handle a larger number of http requests, so> that preference allows a WebDNA admin to adjust the thread limit> accordingly.> >> -----Original Message----->> From: WebCatalog Talk>> [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf>> Of Alain Russell>> Sent: Monday, November 04, 2002 7:38 PM>> To: WebCatalog Talk>> Subject: Re: WebCatalog Eating 200% of the CPU>> >> >> Scott,>> >> I disagree here a bit - If I can bring down another persons>> server that is>> running code that is 'clean' then surely that sets us all up for DOS>> attacks?>> >> If the engine running in multithreaded mode starts so many>> threads that it>> goes into a zombie state and does not respond to any requests>> why have the>> preference?>> >> Were not talking about bringing the engine to a crawl - when>> I simulate this>> the engine stops responding totally. The machine it is running on gets>> unresponsive and a lot of the time the only way to fix it is>> hard reset.>> >> Alain>> >> >> >>> This is not necessarily a bug with the engine. If you>> intentionally (or>>> accidentally) create a template that performs 10s of>> thousands of searches,>>> or that contains a huge number of iterations on a block of>> WebDNA code, to>>> the point where it takes several minutes for the template>> to complete; then>>> if that template is requested dozens of times within a few>> seconds, it would>>> be easy to bring the engine to a crawl. This is why it is>> important for the>>> WebDNA programmer (or any programmer) to exercise some>> caution with their>>> code. It is always a good idea lookout for possible>> infinite or very large>>> loops, redundant code, assumptions in form data, etc...>>> >>> It would be easy to write a C++ application that>> intentionally brings an OS>>> to its knees, so would that be the fault of the OS?>>> >>> However, there are a couple of preferences that may help to>> prevent a WebDNA>>> overload. The 'Max Search Results' was intended to prevent>> an accidental>>> iteration of a large number of search results in the>> [founditems] loop. We>>> have also added a new pref that allows you to control the>> maximum number of>>> concurrent threads the WebDNA engine can spawn. Both of>> those may help.>>> But again, there is no substitute for thorough testing of>> WebDNA template>>> code before going to production.>>> >>> >>> >>>> -----Original Message----->>>> From: WebCatalog Talk>>>> [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf>>>> Of Alain Russell>>>> Sent: Monday, November 04, 2002 6:00 PM>>>> To: WebCatalog Talk>>>> Subject: Re: WebCatalog Eating 200% of the CPU>>>> >>>> >>>> No I think it is supposed to catch the 'Zombie' state ->>>> perhaps a little>>>> more work is needed ?? Perhaps this is why the 4.5 patch>>>> seems to have been>>>> delayed ;)>>>> >>>>> On 11/4/02 4:37 PM, Alain Russell>>>>
wrote:>>>>> >>>>>> Aaron & I had a discussion about this the other day.>>>>>> I've reported to SMSI a while ago about WebDNA doing this>>>> with 'good'>>>>>> searches when the pages are loaded ..>>>>>> >>>>>> Say you have a slow/big search that is coded right - hit>>>> is a few times>>>>>> quickly and this condition can happen.>>>>>> >>>>>> Check the archives for the discussion.>>>>> >>>>> Any response from Smith Micro? I thought WebCatalog>>>> Monitor was supposed to>>>>> catch these situations, or is that only when WebCatalog quits?>>>>> >>>>> Thanks>>>> >>>> >>>> ------------------------------------------------------------->>>> This message is sent to you because you are subscribed to>>>> the mailing list .>>>> To unsubscribe, E-mail to: >>>> To switch to the DIGEST mode, E-mail to>>>> >>>> Web Archive of this list is at: http://search.smithmicro.com/>>> >>> ------------------------------------------------------------->>> This message is sent to you because you are subscribed to>>> the mailing list .>>> To unsubscribe, E-mail to: >>> To switch to the DIGEST mode, E-mail to>>> >>> Web Archive of this list is at: http://search.smithmicro.com/>> >> >> ------------------------------------------------------------->> This message is sent to you because you are subscribed to>> the mailing list .>> To unsubscribe, E-mail to: >> To switch to the DIGEST mode, E-mail to>> >> Web Archive of this list is at: http://search.smithmicro.com/> > -------------------------------------------------------------> This message is sent to you because you are subscribed to> the mailing list .> To unsubscribe, E-mail to: > To switch to the DIGEST mode, E-mail to> > Web Archive of this list is at: http://search.smithmicro.com/> -------------------------------------------------------------This message is sent to you because you are subscribed to the mailing list .To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to Web Archive of this list is at: http://search.smithmicro.com/
Associated Messages, from the most recent to the oldest:
Are you talking about exclusivelock or are there other 'tricks' as well?On 11/5/02 11:11 AM, Scott Anderson mashed the following keys :> What do you mean by 'clean'? A template may contain valid WebDNA syntax,> but could still take a long time to process. I would consider such a> template 'not' a very good candidate for a generally accessible WebDNA> template.> > Your argument assumes that you (or a hacker) have knowledge of a particular> template (or were able to create a template) on the server that can tie up> the engine enough to overwhelm it by sending several http requests for that> template. If no such template existed, then there is not a risk. Again,> this boils down to how carefully and efficiently you have programmed your> templates.> > If it is necessary to create a WebDNA template that will take a large amount> of time to complete (often some admin process), then you should put the> necessary WebDNA code in place to synchronize access to that template to> ensure that only one instance of that template will run at any given time.> Templates that you serve to the general public should be 'lean and mean' and> not 'over loaded' with to much code.> >> why have the preference?> A high performance server can handle a larger number of http requests, so> that preference allows a WebDNA admin to adjust the thread limit> accordingly.> >> -----Original Message----->> From: WebCatalog Talk>> [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf>> Of Alain Russell>> Sent: Monday, November 04, 2002 7:38 PM>> To: WebCatalog Talk>> Subject: Re: WebCatalog Eating 200% of the CPU>> >> >> Scott,>> >> I disagree here a bit - If I can bring down another persons>> server that is>> running code that is 'clean' then surely that sets us all up for DOS>> attacks?>> >> If the engine running in multithreaded mode starts so many>> threads that it>> goes into a zombie state and does not respond to any requests>> why have the>> preference?>> >> Were not talking about bringing the engine to a crawl - when>> I simulate this>> the engine stops responding totally. The machine it is running on gets>> unresponsive and a lot of the time the only way to fix it is>> hard reset.>> >> Alain>> >> >> >>> This is not necessarily a bug with the engine. If you>> intentionally (or>>> accidentally) create a template that performs 10s of>> thousands of searches,>>> or that contains a huge number of iterations on a block of>> WebDNA code, to>>> the point where it takes several minutes for the template>> to complete; then>>> if that template is requested dozens of times within a few>> seconds, it would>>> be easy to bring the engine to a crawl. This is why it is>> important for the>>> WebDNA programmer (or any programmer) to exercise some>> caution with their>>> code. It is always a good idea lookout for possible>> infinite or very large>>> loops, redundant code, assumptions in form data, etc...>>> >>> It would be easy to write a C++ application that>> intentionally brings an OS>>> to its knees, so would that be the fault of the OS?>>> >>> However, there are a couple of preferences that may help to>> prevent a WebDNA>>> overload. The 'Max Search Results' was intended to prevent>> an accidental>>> iteration of a large number of search results in the>> [founditems] loop. We>>> have also added a new pref that allows you to control the>> maximum number of>>> concurrent threads the WebDNA engine can spawn. Both of>> those may help.>>> But again, there is no substitute for thorough testing of>> WebDNA template>>> code before going to production.>>> >>> >>> >>>> -----Original Message----->>>> From: WebCatalog Talk>>>> [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf>>>> Of Alain Russell>>>> Sent: Monday, November 04, 2002 6:00 PM>>>> To: WebCatalog Talk>>>> Subject: Re: WebCatalog Eating 200% of the CPU>>>> >>>> >>>> No I think it is supposed to catch the 'Zombie' state ->>>> perhaps a little>>>> more work is needed ?? Perhaps this is why the 4.5 patch>>>> seems to have been>>>> delayed ;)>>>> >>>>> On 11/4/02 4:37 PM, Alain Russell>>>> wrote:>>>>> >>>>>> Aaron & I had a discussion about this the other day.>>>>>> I've reported to SMSI a while ago about WebDNA doing this>>>> with 'good'>>>>>> searches when the pages are loaded ..>>>>>> >>>>>> Say you have a slow/big search that is coded right - hit>>>> is a few times>>>>>> quickly and this condition can happen.>>>>>> >>>>>> Check the archives for the discussion.>>>>> >>>>> Any response from Smith Micro? I thought WebCatalog>>>> Monitor was supposed to>>>>> catch these situations, or is that only when WebCatalog quits?>>>>> >>>>> Thanks>>>> >>>> >>>> ------------------------------------------------------------->>>> This message is sent to you because you are subscribed to>>>> the mailing list .>>>> To unsubscribe, E-mail to: >>>> To switch to the DIGEST mode, E-mail to>>>> >>>> Web Archive of this list is at: http://search.smithmicro.com/>>> >>> ------------------------------------------------------------->>> This message is sent to you because you are subscribed to>>> the mailing list .>>> To unsubscribe, E-mail to: >>> To switch to the DIGEST mode, E-mail to>>> >>> Web Archive of this list is at: http://search.smithmicro.com/>> >> >> ------------------------------------------------------------->> This message is sent to you because you are subscribed to>> the mailing list .>> To unsubscribe, E-mail to: >> To switch to the DIGEST mode, E-mail to>> >> Web Archive of this list is at: http://search.smithmicro.com/> > -------------------------------------------------------------> This message is sent to you because you are subscribed to> the mailing list .> To unsubscribe, E-mail to: > To switch to the DIGEST mode, E-mail to> > Web Archive of this list is at: http://search.smithmicro.com/> -------------------------------------------------------------This message is sent to you because you are subscribed to the mailing list .To unsubscribe, E-mail to: To switch to the DIGEST mode, E-mail to Web Archive of this list is at: http://search.smithmicro.com/
Aaron Lynch
DOWNLOAD WEBDNA NOW!
Top Articles:
Talk List
The WebDNA community talk-list is the best place to get some help: several hundred extremely proficient programmers with an excellent knowledge of WebDNA and an excellent spirit will deliver all the tips and tricks you can imagine...
Related Readings:
[WebDNA] More YouTube Comments (2010)
The BUG is BACK -- leading < character (1998)
Dark Horse Comics success story (1997)
WebCatalog as a ListServ (1998)
OT: anyone know how to prevent image hijacking with Apache? (2002)
Looking up two prices in database? (1997)
Emailer problems (1998)
3 fields must match (2004)
Triggers under MacOSx (2003)
Separate SSL Server (1997)
WebCatalog can't find database (1997)
searching for items that begin with a number (2004)
Hide Ifs (2000)
[WebDNA] Is anybody running a CAPTCHA? (2017)
Multiple prices (1997)
Problem with a search context (2005)
WebCatalog/Mac 2.1b2 - PIXO (1997)
SKU (1997)
Followup question re: checksum for credit cards? (1997)
(2000)