Re: CFMX Server Crash #4 for the day..
On Wednesday, Nov 27, 2002, at 14:45 US/Pacific, Brook Davies wrote: What is Macromedia doing about this? CFMX is not ready for production. Well, there's quite a few successful production sites out there running CFMX that would beg to differ with you. Our own ColdFusion Examples server - hosting extended documentation and live examples - has been running unattended since June (when we upgraded it to the release version). It had been running the beta successfully since March. I even posted the traffic figures here at one point to show if was coping with quite a bit of traffic. I certainly, based on my experience, would not recommend (and I think anyone running CFMX would agree) anyone to switch to CFMX anytime soon. Unless you have plenty of spare time to baby sit your server. We have servers running CFMX that require no baby-sitting at all and can support very high traffic. Have you contacted Macromedia Technical Support to engage someone to assist you resolving the - obviously serious - problems you are having? Your experience is, I think, fairly unusual (although I see Lee had possibly similar problems - Macromedia worked with him directly on those issues). Sean A Corfield -- Director, Architecture Web Technology Group -- Macromedia, Inc. tel: (415) 252-2287 -- cell: (415) 717-8473 aim: seancorfield -- http://www.macromedia.com An Architect's View -- http://www.corfield.org/blog/ Introducing Macromedia Contribute. Web publishing for everyone. Learn more at http://www.macromedia.com/contribute ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Structure your ColdFusion code with Fusebox. Get the official book at http://www.fusionauthority.com/bkinfo.cfm
RE: CFMX Server Crash #4 for the day..
Sean, Actually, the problem is not unusual at all. We're one of many who have (on this list and in the forums on MM) been having similar issues. There are several threads going on about it now, and MM's already got at least one person working on the issues. The reality, however, is that it's only under what appear to be certain circumstances that it causes these issues. It appears to be related to SQL, or other DB, connectivity. The problem is, however, that it causes the server to be brought to it's knees, which shouldn't happen. It needs to monitor and manage itself more precisely. And his statement about it not being production ready, in my never-to-be-too-humble opinion, is actually correct. Under load, it performs fine, so long as the problems you see us complaining about don't start to occur. And since it's on different types of systems, with different configurations, it's hard to say that it's a simple bug that is too rare to worry about. (I know you're not saying that.. I'm just pointing that out.) At any rate, this is far more widespread than you might think. I know of at least two other companies that are now moving from CF to other languages due to the instabilities. Several others that refuse to now (and most likely will never) upgrade to CFMX... No matter what happens. You may not have seen all the info that was put out on this list about these problems. However, they're not rare.. unfortunately. Lee | -Original Message- | From: Sean A Corfield [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, December 04, 2002 7:58 PM | To: CF-Talk | Subject: Re: CFMX Server Crash #4 for the day.. | | | On Wednesday, Nov 27, 2002, at 14:45 US/Pacific, Brook Davies wrote: | What is Macromedia doing about this? CFMX is not ready for | production. | | Well, there's quite a few successful production sites out | there running | CFMX that would beg to differ with you. Our own ColdFusion Examples | server - hosting extended documentation and live examples - has been | running unattended since June (when we upgraded it to the release | version). It had been running the beta successfully since | March. I even | posted the traffic figures here at one point to show if was | coping with | quite a bit of traffic. | | I certainly, based on my experience, would not recommend | (and I think | anyone running CFMX would agree) anyone to switch to CFMX anytime | soon. Unless you have plenty of spare time to baby sit your server. | | We have servers running CFMX that require no baby-sitting at all and | can support very high traffic. | | Have you contacted Macromedia Technical Support to engage someone to | assist you resolving the - obviously serious - problems you | are having? | Your experience is, I think, fairly unusual (although I see Lee had | possibly similar problems - Macromedia worked with him directly on | those issues). | | Sean A Corfield -- Director, Architecture | Web Technology Group -- Macromedia, Inc. | tel: (415) 252-2287 -- cell: (415) 717-8473 | aim: seancorfield -- http://www.macromedia.com | An Architect's View -- http://www.corfield.org/blog/ | | Introducing Macromedia Contribute. Web publishing for | everyone. Learn more at http://www.macromedia.com/contribute | | ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Get the mailserver that powers this list at http://www.coolfusion.com
Re: CFMX Server Crash #4 for the day..
On Wednesday, Dec 4, 2002, at 20:11 US/Pacific, Lee Fuller wrote: Actually, the problem is not unusual at all. We're one of many who have (on this list and in the forums on MM) been having similar issues. There are several threads going on about it now, and MM's already got at least one person working on the issues. My point was - and I've said this here before - that there are many people successfully running CFMX who do not post. The lists and the forums are where people come when they have problems so the apparent success of CFMX is skewed by what you see here and on the forums. Yes, there are *some* people having problems but there are a lot of successful, happy people using CFMX. That said, your problems are definitely real and we want to address them. Hopefully, the Technical Support folks can work with those customers having problems - like yourself and Brook - and find a suitable resolution. You may not have seen all the info that was put out on this list about these problems. However, they're not rare.. unfortunately. I didn't say rare, I said unusual. The 'norm' is actually success but that's difficult to see based on the posts here and on the forums - and that is perfectly understandable given the nature of these lists and forums. I'm just trying to bring a little balance in response to the claims that CFMX sucks... And it's nice to hear from other users who are successfully using CFMX, especially in production. Sean A Corfield -- Director, Architecture Web Technology Group -- Macromedia, Inc. tel: (415) 252-2287 -- cell: (415) 717-8473 aim: seancorfield -- http://www.macromedia.com An Architect's View -- http://www.corfield.org/blog/ Introducing Macromedia Contribute. Web publishing for everyone. Learn more at http://www.macromedia.com/contribute ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
Re: CFMX Server Crash #4 for the day..
Jochem van Dieten wrote: Another option would be to try and limit the amount of connections to the database to something less as the restart threshold. Then you would start to see timeouts from queries not being able to get a database connection before the server actually crashes. To answer this one myself: limit database connections does not function correctly. I set it to 3, point the browser to a URL where the query takes a few seconds, keep my finger on the refresh button for a while, and suddenly I have 10 database connections running. If I understand the model correctly, connection pooling is implemented in the J2EE Application Server, so that would imply a bug in JRUN. The scary thing is that everybody who is running a shared database with limited client access licenses is easily DoS'ed because 1 customer can eat all the licenses. And the only thing you can do about it is lower the number of simultaneous running threads, because that will limit the number of database connections, but that simply moves the point that gets DoS'ed from the database server to the CF MX server. Jochem ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Your ad could be here. Monies from ads go to support these lists and provide more resources for the community. http://www.fusionauthority.com/ads.cfm
RE: CFMX Server Crash #4 for the day..
Brook, below is a suggestion passed on to me: -- Since they are using SQL the timeout attribute is supported in cfquery, it's not in oracle. They should add a timeout to all their queries. That way if SQL is caught in a deadlock situation SQL will use that timeout attribute as a timer, and if the query waits that long SQL will throw and error back to CF, and more important it will hand the thread back to CF so it can continue handling request. -- have you tried this? Also, have you contacted support? I don't want to restart the whole support thread, but you should get a little more prompt / escelated support from them. mike chambers [EMAIL PROTECTED] -Original Message- From: Brook Davies [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 27, 2002 6:20 PM To: CF-Talk Subject: Re: CFMX Server Crash #4 for the day.. I wish it was, but we have only one app on the server, and it does not use session variables and the application var usage has been thoroughly reviewed and the locks appear correct. The application vars used by the app are refreshed essentially once a day or when they do not exist (isdefined), these locks would not have been in effect at the time of these crashed. The application vars would have been being read but not written. This is the case throughout the site with no exceptions. There are several 3rd party calls, and I had thought initially that this was the problem. But I am not alone with this problem as it seems Tyler is with his Success stories. Tyler, do you think every one in this thread has some application issues and that we should all stop blaming CFMX and use your limited success as proof that CFMX works under load? http://webforums.macromedia.com/coldfusion/messageview.cfm?cat id=143threadi d=468954 Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq This list and all House of Fusion resources hosted by CFHosting.com. The place for dependable ColdFusion Hosting.
RE: CFMX Server Crash #4 for the day..
I'll try this today and see if it helps - thank you. At 10:35 AM 28/11/02 -0500, you wrote: Brook, below is a suggestion passed on to me: -- Since they are using SQL the timeout attribute is supported in cfquery, it's not in oracle. They should add a timeout to all their queries. That way if SQL is caught in a deadlock situation SQL will use that timeout attribute as a timer, and if the query waits that long SQL will throw and error back to CF, and more important it will hand the thread back to CF so it can continue handling request. -- have you tried this? Also, have you contacted support? I don't want to restart the whole support thread, but you should get a little more prompt / escelated support from them. mike chambers [EMAIL PROTECTED] -Original Message- From: Brook Davies [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 27, 2002 6:20 PM To: CF-Talk Subject: Re: CFMX Server Crash #4 for the day.. I wish it was, but we have only one app on the server, and it does not use session variables and the application var usage has been thoroughly reviewed and the locks appear correct. The application vars used by the app are refreshed essentially once a day or when they do not exist (isdefined), these locks would not have been in effect at the time of these crashed. The application vars would have been being read but not written. This is the case throughout the site with no exceptions. There are several 3rd party calls, and I had thought initially that this was the problem. But I am not alone with this problem as it seems Tyler is with his Success stories. Tyler, do you think every one in this thread has some application issues and that we should all stop blaming CFMX and use your limited success as proof that CFMX works under load? http://webforums.macromedia.com/coldfusion/messageview.cfm?cat id=143threadi d=468954 Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Your ad could be here. Monies from ads go to support these lists and provide more resources for the community. http://www.fusionauthority.com/ads.cfm
Re: CFMX Server Crash #4 for the day..
Mike Chambers wrote: -- Since they are using SQL the timeout attribute is supported in cfquery, it's not in oracle. They should add a timeout to all their queries. That way if SQL is caught in a deadlock situation SQL will use that timeout attribute as a timer, and if the query waits that long SQL will throw and error back to CF, and more important it will hand the thread back to CF so it can continue handling request. -- And how would one do that for the queries for client variables (which are handled by CF in the background)? Jochem ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
Re: CFMX Server Crash #4 for the day..
Ahh, very good question/observation. Because , I have seen a page with no queries cause hung threads, and since we are using client var / database storage, this could be the culprit. At 04:40 PM 28/11/02 +0100, you wrote: Mike Chambers wrote: -- Since they are using SQL the timeout attribute is supported in cfquery, it's not in oracle. They should add a timeout to all their queries. That way if SQL is caught in a deadlock situation SQL will use that timeout attribute as a timer, and if the query waits that long SQL will throw and error back to CF, and more important it will hand the thread back to CF so it can continue handling request. -- And how would one do that for the queries for client variables (which are handled by CF in the background)? Jochem ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
RE: CFMX Server Crash #4 for the day..
Mike, We have been working with Mike Collins on this. He has COMPLETE access to one of our servers that has been experiencing these problems, and is quite familiar with how we have things setup there. Just FYI. Lee | -Original Message- | From: Mike Chambers [mailto:[EMAIL PROTECTED]] | Sent: Thursday, November 28, 2002 7:35 AM | To: CF-Talk | Subject: RE: CFMX Server Crash #4 for the day.. | | | Brook, | | below is a suggestion passed on to me: | | -- | Since they are using SQL the timeout attribute is supported | in cfquery, it's not in oracle. They should add a timeout | to all their queries. That way if SQL is caught in a deadlock | situation SQL will use that timeout attribute as a timer, and | if the query waits that long SQL will throw and error back to | CF, and more important it will hand the thread back to CF so | it can continue handling request. | -- | | have you tried this? | | Also, have you contacted support? I don't want to restart the | whole support thread, but you should get a little more prompt | / escelated support from them. | | mike chambers | | [EMAIL PROTECTED] | | | -Original Message- | From: Brook Davies [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 6:20 PM | To: CF-Talk | Subject: Re: CFMX Server Crash #4 for the day.. | | | I wish it was, but we have only one app on the server, and it | does not use | session variables and the application var usage has been thoroughly | reviewed and the locks appear correct. | | The application vars used by the app are refreshed | essentially once a day | or when they do not exist (isdefined), these locks would not | have been in | effect at the time of these crashed. The application vars | would have been | being read but not written. This is the case throughout the | site with no | exceptions. | | There are several 3rd party calls, and I had thought | initially that this | was the problem. But I am not alone with this problem as it | seems Tyler is | with his Success stories. Tyler, do you think every one in | this thread | has some application issues and that we should all stop | blaming CFMX and | use your limited success as proof that CFMX works under load? | | http://webforums.macromedia.com/coldfusion/messageview.cfm?cat | id=143threadi | d=468954 | | Brook | | | ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
RE: CFMX Server Crash #4 for the day..
Exactly. I was just thinking of that after I responded to Mike's comment. REALITY: - CFMX quite simply does not monitor itself and it's connectivity thoroughly enough. - It can be easily killed, with improper programming. - It has less control (via the admin panel) than previous versions, where it comes to dealing with things like, for example, this SQL timeout issue. - They now know about it, and are (more than likely) working hard now to recode some things, add a few features and tweak yet others to make the next SP release handle some of this. soapbox AGAIN: (And I repeat this for the benefit of those who see only my response to one of the many threads on this topic and don't clearly realize where I, and many others, stand on this issue) I cannot stress enough that I am NOT bashing CFMX. I'm simply pointing out reality, in the hope that we can continue to utilize this fantastic product in a true production environment without running the risk of having our servers die every 10 minutes -- WHICH HAS HAPPENED MANY TIMES. /soapbox | And how would one do that for the queries for client variables (which | are handled by CF in the background)? ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq This list and all House of Fusion resources hosted by CFHosting.com. The place for dependable ColdFusion Hosting.
Re: CFMX Server Crash #4 for the day..
Lee Fuller wrote: - It has less control (via the admin panel) than previous versions, where it comes to dealing with things like, for example, this SQL timeout issue. What could you control in previous versions that you can't control anymore? Jochem ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Structure your ColdFusion code with Fusebox. Get the official book at http://www.fusionauthority.com/bkinfo.cfm
RE: CFMX Server Crash #4 for the day..
Actually.. I should rephrase that statement... Thanks Jochem. (I need to remember to be careful when I state things this way -- since it may give the wrong impression.) What I should have said is: It has more features, yet it does not have the control that is should over these new features. Things like monitoring non-dying or out of control threads, etc. Also, vars -- there we a few things in the var controls on previous versions that is not there now. I'd have to go back over my notes to find them. But there were posts about these not too long ago. Sorry for the confusion, and thanks for pointing it out. | -Original Message- | From: Jochem van Dieten [mailto:[EMAIL PROTECTED]] | Sent: Thursday, November 28, 2002 9:57 AM | To: CF-Talk | Subject: Re: CFMX Server Crash #4 for the day.. | | | Lee Fuller wrote: | | - It has less control (via the admin panel) than | previous versions, | where it comes to dealing with things like, for example, this SQL | timeout issue. | | What could you control in previous versions that you can't | control anymore? | | Jochem | | ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Structure your ColdFusion code with Fusebox. Get the official book at http://www.fusionauthority.com/bkinfo.cfm
Re: CFMX Server Crash #4 for the day..
Off the top of my head and perhaps worth just that... Could you have a deadlock condition going? ... Locking AB on one request and another near simultaneous request locking BA such that each are waiting on the other to finish and release the lock so that they can exert their second lock and continue? Just an off the wall thought from previous experiences with other systems. At 02:45 PM 11/27/02, you wrote: Well, our CFMX server just crashed for the 4th time today. I've had the perfmon open all day and have been watching the traffic. We had an average of 12 connections when the running requests started to grow and grow. As I mentioned in my previous post, there were 3 running requests that had not finished for hours. As far as the number of CFMX request, as I was watching it, it never went above 6. This really means there was a max of 3 running requests (on top of the 3 that were already dead), it would go up to 6 and then back down to 3. Those 3 threads weren't going anywhere. Then I noticed that the running requests was growing incrementally, so I tried to make an HTTP request to the serverand it did not respond. I checked the running requests, and there was one more. The server was DEAD, HUNG, FINISHED - KAPUT. The Perfmon counter restarted jrun when the running requests reached 10, and the server came back to life. Thank you Lee Fuller for this temporary fix. What is Macromedia doing about this? CFMX is not ready for production. If your sales of CFMX are not what you expected, there's a fairly obvious reason why. I certainly, based on my experience, would not recommend (and I think anyone running CFMX would agree) anyone to switch to CFMX anytime soon. Unless you have plenty of spare time to baby sit your server. Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Structure your ColdFusion code with Fusebox. Get the official book at http://www.fusionauthority.com/bkinfo.cfm
RE: CFMX Server Crash #4 for the day..
Brook, Since people only seem to mention CFMX when it does bad, I'm going to mention when it does good. I've personally had atleast half dozen CFMX installs go fine. The sites on the boxes run fine. Even under load. So I for one say 'CFMX is ready for production. (just build your site to take advantage of it)'. I would and do recommend it for anyone looking to build a web app. We launched a pretty good site that was half CFMX, half CF5 less than a month after CFMX's release date. No updater was available and we had no problems. The network load-balancing solution our client had selected went down before the CF servers did under load. We had 200 simulateous requests chugging along on our Win2k server with IIS5. If your server is crashing at 6 I'd check your code first and stop complaining about the server. Something bigger seems to be going on. If you upgraded from 5, then did you test your app before hand? Any load testing would have brought this problem up. Good luck, t ** Tyler M. Fitch Certified Advanced ColdFusion 5 Developer http://isitedesign.com ** -Original Message- From: Brook Davies [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 27, 2002 2:45 PM To: CF-Talk Subject: CFMX Server Crash #4 for the day.. Well, our CFMX server just crashed for the 4th time today. I've had the perfmon open all day and have been watching the traffic. We had an average of 12 connections when the running requests started to grow and grow. As I mentioned in my previous post, there were 3 running requests that had not finished for hours. As far as the number of CFMX request, as I was watching it, it never went above 6. This really means there was a max of 3 running requests (on top of the 3 that were already dead), it would go up to 6 and then back down to 3. Those 3 threads weren't going anywhere. Then I noticed that the running requests was growing incrementally, so I tried to make an HTTP request to the serverand it did not respond. I checked the running requests, and there was one more. The server was DEAD, HUNG, FINISHED - KAPUT. The Perfmon counter restarted jrun when the running requests reached 10, and the server came back to life. Thank you Lee Fuller for this temporary fix. What is Macromedia doing about this? CFMX is not ready for production. If your sales of CFMX are not what you expected, there's a fairly obvious reason why. I certainly, based on my experience, would not recommend (and I think anyone running CFMX would agree) anyone to switch to CFMX anytime soon. Unless you have plenty of spare time to baby sit your server. Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Structure your ColdFusion code with Fusebox. Get the official book at http://www.fusionauthority.com/bkinfo.cfm
RE: CFMX Server Crash #4 for the day..
I'm going to guess that your not running SQL Server on a separate box. Are you? At 04:00 PM 27/11/02 -0700, you wrote: Brook, Since people only seem to mention CFMX when it does bad, I'm going to mention when it does good. I've personally had atleast half dozen CFMX installs go fine. The sites on the boxes run fine. Even under load. So I for one say 'CFMX is ready for production. (just build your site to take advantage of it)'. I would and do recommend it for anyone looking to build a web app. We launched a pretty good site that was half CFMX, half CF5 less than a month after CFMX's release date. No updater was available and we had no problems. The network load-balancing solution our client had selected went down before the CF servers did under load. We had 200 simulateous requests chugging along on our Win2k server with IIS5. If your server is crashing at 6 I'd check your code first and stop complaining about the server. Something bigger seems to be going on. If you upgraded from 5, then did you test your app before hand? Any load testing would have brought this problem up. Good luck, t ** Tyler M. Fitch Certified Advanced ColdFusion 5 Developer http://isitedesign.com ** -Original Message- From: Brook Davies [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 27, 2002 2:45 PM To: CF-Talk Subject: CFMX Server Crash #4 for the day.. Well, our CFMX server just crashed for the 4th time today. I've had the perfmon open all day and have been watching the traffic. We had an average of 12 connections when the running requests started to grow and grow. As I mentioned in my previous post, there were 3 running requests that had not finished for hours. As far as the number of CFMX request, as I was watching it, it never went above 6. This really means there was a max of 3 running requests (on top of the 3 that were already dead), it would go up to 6 and then back down to 3. Those 3 threads weren't going anywhere. Then I noticed that the running requests was growing incrementally, so I tried to make an HTTP request to the serverand it did not respond. I checked the running requests, and there was one more. The server was DEAD, HUNG, FINISHED - KAPUT. The Perfmon counter restarted jrun when the running requests reached 10, and the server came back to life. Thank you Lee Fuller for this temporary fix. What is Macromedia doing about this? CFMX is not ready for production. If your sales of CFMX are not what you expected, there's a fairly obvious reason why. I certainly, based on my experience, would not recommend (and I think anyone running CFMX would agree) anyone to switch to CFMX anytime soon. Unless you have plenty of spare time to baby sit your server. Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Structure your ColdFusion code with Fusebox. Get the official book at http://www.fusionauthority.com/bkinfo.cfm
RE: CFMX Server Crash #4 for the day..
We sure are. MSSQL 2k though. That's the only difference I've seen. t ** Tyler M. Fitch Certified Advanced ColdFusion 5 Developer http://isitedesign.com ** -Original Message- From: Brook Davies [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 27, 2002 3:06 PM To: CF-Talk Subject: RE: CFMX Server Crash #4 for the day.. I'm going to guess that your not running SQL Server on a separate box. Are you? At 04:00 PM 27/11/02 -0700, you wrote: Brook, Since people only seem to mention CFMX when it does bad, I'm going to mention when it does good. I've personally had atleast half dozen CFMX installs go fine. The sites on the boxes run fine. Even under load. So I for one say 'CFMX is ready for production. (just build your site to take advantage of it)'. I would and do recommend it for anyone looking to build a web app. We launched a pretty good site that was half CFMX, half CF5 less than a month after CFMX's release date. No updater was available and we had no problems. The network load-balancing solution our client had selected went down before the CF servers did under load. We had 200 simulateous requests chugging along on our Win2k server with IIS5. If your server is crashing at 6 I'd check your code first and stop complaining about the server. Something bigger seems to be going on. If you upgraded from 5, then did you test your app before hand? Any load testing would have brought this problem up. Good luck, t ** Tyler M. Fitch Certified Advanced ColdFusion 5 Developer http://isitedesign.com ** -Original Message- From: Brook Davies [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 27, 2002 2:45 PM To: CF-Talk Subject: CFMX Server Crash #4 for the day.. Well, our CFMX server just crashed for the 4th time today. I've had the perfmon open all day and have been watching the traffic. We had an average of 12 connections when the running requests started to grow and grow. As I mentioned in my previous post, there were 3 running requests that had not finished for hours. As far as the number of CFMX request, as I was watching it, it never went above 6. This really means there was a max of 3 running requests (on top of the 3 that were already dead), it would go up to 6 and then back down to 3. Those 3 threads weren't going anywhere. Then I noticed that the running requests was growing incrementally, so I tried to make an HTTP request to the serverand it did not respond. I checked the running requests, and there was one more. The server was DEAD, HUNG, FINISHED - KAPUT. The Perfmon counter restarted jrun when the running requests reached 10, and the server came back to life. Thank you Lee Fuller for this temporary fix. What is Macromedia doing about this? CFMX is not ready for production. If your sales of CFMX are not what you expected, there's a fairly obvious reason why. I certainly, based on my experience, would not recommend (and I think anyone running CFMX would agree) anyone to switch to CFMX anytime soon. Unless you have plenty of spare time to baby sit your server. Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Get the mailserver that powers this list at http://www.coolfusion.com
RE: CFMX Server Crash #4 for the day..
Brian (and Everyone) We have thoroughly beat this issue up, and found it simply to be that CFMX poorly manages runaway conditions. It's really just that simple. Locking doesn't change it, since we have sites that use absolutely no session, client or other persistent vars, and yet the run out of control. I appears to be DB related -- either due to driver issues/errors, or server discussions. However, in any event, CFMX can (and will -- as we've seen numerous times) run out of control, only to finally become unresponsive. The information I've provided as a perfmon work-around has keep the wolves at bay. But the reality is that they are right outside the door, and any crack lets them in -- with a vengeance. At any rate.. The theory is that the latest SP will fix some of the issues. But in reality, this is truly a design issue -- one that will require somewhat of a redesign on how CFMX services manage themselves, report/handle errors and runaway conditions, etc. As we all know, the Restart Server on X Unresponsive Requests simply has never worked. So here we are... Waiting and hoping for a quick resolution. And again.. DON'T GET ME WRONG -- I TRULY WISH to continue supporting CFMX. I do believe that it is a powerful, flexible and scalable language -- albeit a bit Sponge Bob'ish right now. grin Lee | -Original Message- | From: Brian Scandale [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 2:58 PM | To: CF-Talk | Subject: Re: CFMX Server Crash #4 for the day.. | | | Off the top of my head and perhaps worth just that... | | Could you have a deadlock condition going? ... Locking AB on | one request and another near simultaneous request locking BA | such that each are waiting on the other to finish and release | the lock so that they can exert their second lock and continue? | | Just an off the wall thought from previous experiences with | other systems. | | | | | | At 02:45 PM 11/27/02, you wrote: | Well, our CFMX server just crashed for the 4th time today. | I've had the | perfmon open all day and have been watching the traffic. We | had an average | of 12 connections when the running requests started to grow | and grow. As I | mentioned in my previous post, there were 3 running requests | that had not | finished for hours. As far as the number of CFMX request, as | I was watching | it, it never went above 6. This really means there was a max | of 3 running | requests (on top of the 3 that were already dead), it would | go up to 6 and | then back down to 3. Those 3 threads weren't going anywhere. | | Then I noticed that the running requests was growing | incrementally, so | I | tried to make an HTTP request to the serverand it did | not respond. I | checked the running requests, and there was one more. The | server was DEAD, | HUNG, FINISHED - KAPUT. | | The Perfmon counter restarted jrun when the running requests reached | 10, | and the server came back to life. Thank you Lee Fuller for | this temporary fix. | | What is Macromedia doing about this? CFMX is not ready for | production. | If | your sales of CFMX are not what you expected, there's a | fairly obvious | reason why. I certainly, based on my experience, would not | recommend (and I | think anyone running CFMX would agree) anyone to switch to | CFMX anytime | soon. Unless you have plenty of spare time to baby sit your server. | | Brook | | | | | ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
RE: CFMX Server Crash #4 for the day..
There seems to be something to this issue.. Which we have noted in the thread I mentioned in an earlier post from MM's forum. When SQL (2K or 7) is on the same box, this tends not to happen. However, I recall that we had at least one instance of it occuring regardless of the fact that SQL was on the same box. (Brook, can you confirm/deny this for me? I thought you were involved in that convo.) At any rate, it DOES appear to be DB/DB Driver related. How, specifically, I don't know. And so far, neither does anyone else, from what I have heard/seen/read. If you missed my post of the thread location, here it is: http://webforums.macromedia.com/coldfusion/messageview.cfm?catid=143thr eadid=468954 HTH Lee | -Original Message- | From: Fitch, Tyler [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 3:10 PM | To: CF-Talk | Subject: RE: CFMX Server Crash #4 for the day.. | | | We sure are. MSSQL 2k though. That's the only difference I've seen. | | t | | ** | Tyler M. Fitch | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | ** | | -Original Message- | From: Brook Davies [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 3:06 PM | To: CF-Talk | Subject: RE: CFMX Server Crash #4 for the day.. | | | I'm going to guess that your not running SQL Server on a | separate box. Are you? | | At 04:00 PM 27/11/02 -0700, you wrote: | Brook, | | Since people only seem to mention CFMX when it does bad, I'm going to | mention when it does good. | | I've personally had atleast half dozen CFMX installs go fine. The | sites on the boxes run fine. Even under load. So I for one | say 'CFMX | is ready for production. (just build your site to take advantage of | it)'. I would and do recommend it for anyone looking to build a web | app. | | We launched a pretty good site that was half CFMX, half CF5 | less than a | | month after CFMX's release date. No updater was available and we had | no problems. The network load-balancing solution our client had | selected went down before the CF servers did under load. | | We had 200 simulateous requests chugging along on our Win2k | server with | | IIS5. If your server is crashing at 6 I'd check your code first and | stop complaining about the server. Something bigger seems | to be going | on. | | If you upgraded from 5, then did you test your app before hand? Any | load testing would have brought this problem up. | | Good luck, | | t | | * | * | Tyler M. Fitch | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | * | * | | -Original Message- | From: Brook Davies [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 2:45 PM | To: CF-Talk | Subject: CFMX Server Crash #4 for the day.. | | | Well, our CFMX server just crashed for the 4th time today. | I've had the | | perfmon open all day and have been watching the traffic. We had an | average of 12 connections when the running requests started | to grow and | | grow. As I | mentioned in my previous post, there were 3 running requests | that had | not finished for hours. As far as the number of CFMX | request, as I was | watching | it, it never went above 6. This really means there was a max of 3 | running | requests (on top of the 3 that were already dead), it would | go up to 6 | and | then back down to 3. Those 3 threads weren't going anywhere. | | Then I noticed that the running requests was growing | incrementally, so | I | | tried to make an HTTP request to the serverand it did | not respond. | I | | checked the running requests, and there was one more. The server was | DEAD, HUNG, FINISHED - KAPUT. | | The Perfmon counter restarted jrun when the running requests reached | 10, | | and the server came back to life. Thank you Lee Fuller for this | temporary fix. | | What is Macromedia doing about this? CFMX is not ready for | production. | If your sales of CFMX are not what you expected, there's a fairly | obvious reason why. I certainly, based on my experience, would not | recommend (and I | think anyone running CFMX would agree) anyone to switch to | CFMX anytime | soon. Unless you have plenty of spare time to baby sit your server. | | Brook | | | | | | | ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Your ad could be here. Monies from ads go to support these lists and provide more resources for the community. http://www.fusionauthority.com/ads.cfm
Re: CFMX Server Crash #4 for the day..
I wish it was, but we have only one app on the server, and it does not use session variables and the application var usage has been thoroughly reviewed and the locks appear correct. The application vars used by the app are refreshed essentially once a day or when they do not exist (isdefined), these locks would not have been in effect at the time of these crashed. The application vars would have been being read but not written. This is the case throughout the site with no exceptions. There are several 3rd party calls, and I had thought initially that this was the problem. But I am not alone with this problem as it seems Tyler is with his Success stories. Tyler, do you think every one in this thread has some application issues and that we should all stop blaming CFMX and use your limited success as proof that CFMX works under load? http://webforums.macromedia.com/coldfusion/messageview.cfm?catid=143threadi d=468954 Brook At 02:58 PM 27/11/02 -0800, you wrote: Off the top of my head and perhaps worth just that... Could you have a deadlock condition going? ... Locking AB on one request and another near simultaneous request locking BA such that each are waiting on the other to finish and release the lock so that they can exert their second lock and continue? Just an off the wall thought from previous experiences with other systems. At 02:45 PM 11/27/02, you wrote: Well, our CFMX server just crashed for the 4th time today. I've had the perfmon open all day and have been watching the traffic. We had an average of 12 connections when the running requests started to grow and grow. As I mentioned in my previous post, there were 3 running requests that had not finished for hours. As far as the number of CFMX request, as I was watching it, it never went above 6. This really means there was a max of 3 running requests (on top of the 3 that were already dead), it would go up to 6 and then back down to 3. Those 3 threads weren't going anywhere. Then I noticed that the running requests was growing incrementally, so I tried to make an HTTP request to the serverand it did not respond. I checked the running requests, and there was one more. The server was DEAD, HUNG, FINISHED - KAPUT. The Perfmon counter restarted jrun when the running requests reached 10, and the server came back to life. Thank you Lee Fuller for this temporary fix. What is Macromedia doing about this? CFMX is not ready for production. If your sales of CFMX are not what you expected, there's a fairly obvious reason why. I certainly, based on my experience, would not recommend (and I think anyone running CFMX would agree) anyone to switch to CFMX anytime soon. Unless you have plenty of spare time to baby sit your server. Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
RE: CFMX Server Crash #4 for the day..
I agree with this completely and have voiced my support for CFMX in the past. Other than this issue, I think CFMX is an incredible achievement by MM. I think they have moved it in a very positive direction. But when your server is crashing 4 times in a day, IMHO, I have a right to be upset. Brook At 03:14 PM 27/11/02 -0800, you wrote: So here we are... Waiting and hoping for a quick resolution. And again.. DON'T GET ME WRONG -- I TRULY WISH to continue supporting CFMX. I do believe that it is a powerful, flexible and scalable language -- albeit a bit Sponge Bob'ish right now. grin ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Your ad could be here. Monies from ads go to support these lists and provide more resources for the community. http://www.fusionauthority.com/ads.cfm
RE: CFMX Server Crash #4 for the day..
It might help if you posted your configuration, if I were you and were experiencing this I'd probably call support though. Just my $.02 Rob -Original Message- From: Brook Davies [mailto:[EMAIL PROTECTED]] Sent: Wednesday, November 27, 2002 5:45 PM To: CF-Talk Subject: CFMX Server Crash #4 for the day.. Well, our CFMX server just crashed for the 4th time today. I've had the perfmon open all day and have been watching the traffic. We had an average of 12 connections when the running requests started to grow and grow. As I mentioned in my previous post, there were 3 running requests that had not finished for hours. As far as the number of CFMX request, as I was watching it, it never went above 6. This really means there was a max of 3 running requests (on top of the 3 that were already dead), it would go up to 6 and then back down to 3. Those 3 threads weren't going anywhere. Then I noticed that the running requests was growing incrementally, so I tried to make an HTTP request to the serverand it did not respond. I checked the running requests, and there was one more. The server was DEAD, HUNG, FINISHED - KAPUT. The Perfmon counter restarted jrun when the running requests reached 10, and the server came back to life. Thank you Lee Fuller for this temporary fix. What is Macromedia doing about this? CFMX is not ready for production. If your sales of CFMX are not what you expected, there's a fairly obvious reason why. I certainly, based on my experience, would not recommend (and I think anyone running CFMX would agree) anyone to switch to CFMX anytime soon. Unless you have plenty of spare time to baby sit your server. Brook ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
RE: CFMX Server Crash #4 for the day..
I think there are a couple of different ways to connect to the MSSQL engines... Socket connections for example and a few flavors of drivers I believe. If I am correct you might try them to see what effect it has. I am connected to SQL2000 via a socket driver I think it was called. for what its worth ;-( At 03:17 PM 11/27/02, you wrote: There seems to be something to this issue.. Which we have noted in the thread I mentioned in an earlier post from MM's forum. When SQL (2K or 7) is on the same box, this tends not to happen. However, I recall that we had at least one instance of it occuring regardless of the fact that SQL was on the same box. (Brook, can you confirm/deny this for me? I thought you were involved in that convo.) At any rate, it DOES appear to be DB/DB Driver related. How, specifically, I don't know. And so far, neither does anyone else, from what I have heard/seen/read. If you missed my post of the thread location, here it is: http://webforums.macromedia.com/coldfusion/messageview.cfm?catid=143thr eadid=468954 HTH Lee | -Original Message- | From: Fitch, Tyler [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 3:10 PM | To: CF-Talk | Subject: RE: CFMX Server Crash #4 for the day.. | | | We sure are. MSSQL 2k though. That's the only difference I've seen. | | t | | ** | Tyler M. Fitch | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | ** | | -Original Message- | From: Brook Davies [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 3:06 PM | To: CF-Talk | Subject: RE: CFMX Server Crash #4 for the day.. | | | I'm going to guess that your not running SQL Server on a | separate box. Are you? | | At 04:00 PM 27/11/02 -0700, you wrote: | Brook, | | Since people only seem to mention CFMX when it does bad, I'm going to | mention when it does good. | | I've personally had atleast half dozen CFMX installs go fine. The | sites on the boxes run fine. Even under load. So I for one | say 'CFMX | is ready for production. (just build your site to take advantage of | it)'. I would and do recommend it for anyone looking to build a web | app. | | We launched a pretty good site that was half CFMX, half CF5 | less than a | | month after CFMX's release date. No updater was available and we had | no problems. The network load-balancing solution our client had | selected went down before the CF servers did under load. | | We had 200 simulateous requests chugging along on our Win2k | server with | | IIS5. If your server is crashing at 6 I'd check your code first and | stop complaining about the server. Something bigger seems | to be going | on. | | If you upgraded from 5, then did you test your app before hand? Any | load testing would have brought this problem up. | | Good luck, | | t | | * | * | Tyler M. Fitch | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | * | * | | -Original Message- | From: Brook Davies [mailto:[EMAIL PROTECTED]] | Sent: Wednesday, November 27, 2002 2:45 PM | To: CF-Talk | Subject: CFMX Server Crash #4 for the day.. | | | Well, our CFMX server just crashed for the 4th time today. | I've had the | | perfmon open all day and have been watching the traffic. We had an | average of 12 connections when the running requests started | to grow and | | grow. As I | mentioned in my previous post, there were 3 running requests | that had | not finished for hours. As far as the number of CFMX | request, as I was | watching | it, it never went above 6. This really means there was a max of 3 | running | requests (on top of the 3 that were already dead), it would | go up to 6 | and | then back down to 3. Those 3 threads weren't going anywhere. | | Then I noticed that the running requests was growing | incrementally, so | I | | tried to make an HTTP request to the serverand it did | not respond. | I | | checked the running requests, and there was one more. The server was | DEAD, HUNG, FINISHED - KAPUT. | | The Perfmon counter restarted jrun when the running requests reached | 10, | | and the server came back to life. Thank you Lee Fuller for this | temporary fix. | | What is Macromedia doing about this? CFMX is not ready for | production. | If your sales of CFMX are not what you expected, there's a fairly | obvious reason why. I certainly, based on my experience, would not | recommend (and I | think anyone running CFMX would agree) anyone to switch to | CFMX anytime | soon. Unless you have plenty of spare time to baby sit your server. | | Brook | | | | | | | ~| Archives: http
Re: CFMX Server Crash #4 for the day..
Don't tell my machine then, I run CFMX on RedHat Linux, and use an ODBC connection to SQL2000 on a Windows box - no problems so far. = Douglas White group Manager mailto:[EMAIL PROTECTED] http://www.samcfug.org = - Original Message - From: Lee Fuller [EMAIL PROTECTED] To: CF-Talk [EMAIL PROTECTED] Sent: Wednesday, November 27, 2002 5:17 PM Subject: RE: CFMX Server Crash #4 for the day.. | There seems to be something to this issue.. Which we have noted in the | thread I mentioned in an earlier post from MM's forum. When SQL (2K or | 7) is on the same box, this tends not to happen. However, I recall that | we had at least one instance of it occuring regardless of the fact that | SQL was on the same box. (Brook, can you confirm/deny this for me? I | thought you were involved in that convo.) At any rate, it DOES appear | to be DB/DB Driver related. How, specifically, I don't know. And so | far, neither does anyone else, from what I have heard/seen/read. | | If you missed my post of the thread location, here it is: | | | http://webforums.macromedia.com/coldfusion/messageview.cfm?catid=143thr | eadid=468954 | | HTH | | Lee | | | | -Original Message- | | From: Fitch, Tyler [mailto:[EMAIL PROTECTED]] | | Sent: Wednesday, November 27, 2002 3:10 PM | | To: CF-Talk | | Subject: RE: CFMX Server Crash #4 for the day.. | | | | | | We sure are. MSSQL 2k though. That's the only difference I've seen. | | | | t | | | | ** | | Tyler M. Fitch | | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | | ** | | | | -Original Message- | | From: Brook Davies [mailto:[EMAIL PROTECTED]] | | Sent: Wednesday, November 27, 2002 3:06 PM | | To: CF-Talk | | Subject: RE: CFMX Server Crash #4 for the day.. | | | | | | I'm going to guess that your not running SQL Server on a | | separate box. Are you? | | | | At 04:00 PM 27/11/02 -0700, you wrote: | | Brook, | | | | Since people only seem to mention CFMX when it does bad, I'm going to | | mention when it does good. | | | | I've personally had atleast half dozen CFMX installs go fine. The | | sites on the boxes run fine. Even under load. So I for one | | say 'CFMX | | is ready for production. (just build your site to take advantage of | | it)'. I would and do recommend it for anyone looking to build a web | | app. | | | | We launched a pretty good site that was half CFMX, half CF5 | | less than a | | | | month after CFMX's release date. No updater was available and we had | | no problems. The network load-balancing solution our client had | | selected went down before the CF servers did under load. | | | | We had 200 simulateous requests chugging along on our Win2k | | server with | | | | IIS5. If your server is crashing at 6 I'd check your code first and | | stop complaining about the server. Something bigger seems | | to be going | | on. | | | | If you upgraded from 5, then did you test your app before hand? Any | | load testing would have brought this problem up. | | | | Good luck, | | | | t | | | | * | | * | | Tyler M. Fitch | | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | | * | | * | | | | -Original Message- | | From: Brook Davies [mailto:[EMAIL PROTECTED]] | | Sent: Wednesday, November 27, 2002 2:45 PM | | To: CF-Talk | | Subject: CFMX Server Crash #4 for the day.. | | | | | | Well, our CFMX server just crashed for the 4th time today. | | I've had the | | | | perfmon open all day and have been watching the traffic. We had an | | average of 12 connections when the running requests started | | to grow and | | | | grow. As I | | mentioned in my previous post, there were 3 running requests | | that had | | not finished for hours. As far as the number of CFMX | | request, as I was | | watching | | it, it never went above 6. This really means there was a max of 3 | | running | | requests (on top of the 3 that were already dead), it would | | go up to 6 | | and | | then back down to 3. Those 3 threads weren't going anywhere. | | | | Then I noticed that the running requests was growing | | incrementally, so | | I | | | | tried to make an HTTP request to the serverand it did | | not respond. | | I | | | | checked the running requests, and there was one more. The server was | | DEAD, HUNG, FINISHED - KAPUT. | | | | The Perfmon counter restarted jrun when the running requests reached | | 10, | | | | and the server came back to life. Thank you Lee Fuller for this | | temporary fix. | | | | What is Macromedia doing about this? CFMX is not ready for | | production. | | If your sales of CFMX are not what you expected, there's a fairly | | obvious reason why
Re: CFMX Server Crash #4 for the day..
We won't tell it if you don't.. At 05:52 PM 27/11/02 -0600, you wrote: Don't tell my machine then, I run CFMX on RedHat Linux, and use an ODBC connection to SQL2000 on a Windows box - no problems so far. = Douglas White group Manager mailto:[EMAIL PROTECTED] http://www.samcfug.org = - Original Message - From: Lee Fuller [EMAIL PROTECTED] To: CF-Talk [EMAIL PROTECTED] Sent: Wednesday, November 27, 2002 5:17 PM Subject: RE: CFMX Server Crash #4 for the day.. | There seems to be something to this issue.. Which we have noted in the | thread I mentioned in an earlier post from MM's forum. When SQL (2K or | 7) is on the same box, this tends not to happen. However, I recall that | we had at least one instance of it occuring regardless of the fact that | SQL was on the same box. (Brook, can you confirm/deny this for me? I | thought you were involved in that convo.) At any rate, it DOES appear | to be DB/DB Driver related. How, specifically, I don't know. And so | far, neither does anyone else, from what I have heard/seen/read. | | If you missed my post of the thread location, here it is: | | | http://webforums.macromedia.com/coldfusion/messageview.cfm?catid=143thr | eadid=468954 | | HTH | | Lee | | | | -Original Message- | | From: Fitch, Tyler [mailto:[EMAIL PROTECTED]] | | Sent: Wednesday, November 27, 2002 3:10 PM | | To: CF-Talk | | Subject: RE: CFMX Server Crash #4 for the day.. | | | | | | We sure are. MSSQL 2k though. That's the only difference I've seen. | | | | t | | | | ** | | Tyler M. Fitch | | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | | ** | | | | -Original Message- | | From: Brook Davies [mailto:[EMAIL PROTECTED]] | | Sent: Wednesday, November 27, 2002 3:06 PM | | To: CF-Talk | | Subject: RE: CFMX Server Crash #4 for the day.. | | | | | | I'm going to guess that your not running SQL Server on a | | separate box. Are you? | | | | At 04:00 PM 27/11/02 -0700, you wrote: | | Brook, | | | | Since people only seem to mention CFMX when it does bad, I'm going to | | mention when it does good. | | | | I've personally had atleast half dozen CFMX installs go fine. The | | sites on the boxes run fine. Even under load. So I for one | | say 'CFMX | | is ready for production. (just build your site to take advantage of | | it)'. I would and do recommend it for anyone looking to build a web | | app. | | | | We launched a pretty good site that was half CFMX, half CF5 | | less than a | | | | month after CFMX's release date. No updater was available and we had | | no problems. The network load-balancing solution our client had | | selected went down before the CF servers did under load. | | | | We had 200 simulateous requests chugging along on our Win2k | | server with | | | | IIS5. If your server is crashing at 6 I'd check your code first and | | stop complaining about the server. Something bigger seems | | to be going | | on. | | | | If you upgraded from 5, then did you test your app before hand? Any | | load testing would have brought this problem up. | | | | Good luck, | | | | t | | | | * | | * | | Tyler M. Fitch | | Certified Advanced ColdFusion 5 Developer http://isitedesign.com | | * | | * | | | | -Original Message- | | From: Brook Davies [mailto:[EMAIL PROTECTED]] | | Sent: Wednesday, November 27, 2002 2:45 PM | | To: CF-Talk | | Subject: CFMX Server Crash #4 for the day.. | | | | | | Well, our CFMX server just crashed for the 4th time today. | | I've had the | | | | perfmon open all day and have been watching the traffic. We had an | | average of 12 connections when the running requests started | | to grow and | | | | grow. As I | | mentioned in my previous post, there were 3 running requests | | that had | | not finished for hours. As far as the number of CFMX | | request, as I was | | watching | | it, it never went above 6. This really means there was a max of 3 | | running | | requests (on top of the 3 that were already dead), it would | | go up to 6 | | and | | then back down to 3. Those 3 threads weren't going anywhere. | | | | Then I noticed that the running requests was growing | | incrementally, so | | I | | | | tried to make an HTTP request to the serverand it did | | not respond. | | I | | | | checked the running requests, and there was one more. The server was | | DEAD, HUNG, FINISHED - KAPUT. | | | | The Perfmon counter restarted jrun when the running requests reached | | 10, | | | | and the server came back to life. Thank you Lee Fuller for this | | temporary fix. | | | | What is Macromedia doing about this? CFMX is not ready for | | production. | | If your sales
Re: CFMX Server Crash #4 for the day..
Lee Fuller wrote: I appears to be DB related -- either due to driver issues/errors, or server discussions. However, in any event, CFMX can (and will -- as we've seen numerous times) run out of control, only to finally become unresponsive. Piling up some assumptions: First of all, CF MX will wait indefinitely for database queries to return if there is a problem in the JDBC driver or the database that does not return an error condition. This will add to the running request number as well as eat 1 client access licence from the dataabse server. This means that following request will have less threads to work with, and have to open a new databse connection. Eventually, this database connection also breaks. Next client access license + thread eaten. Process repeats itself until the maximum number of allowed requests is reached. Some way to test this scenario would be to disable connection pooling. Then you should be able to see on the database side how long connections stay open. If they stay open past the longest query, you found the problem. Another option would be to try and limit the amount of connections to the database to something less as the restart threshold. Then you would start to see timeouts from queries not being able to get a database connection before the server actually crashes. At any rate.. The theory is that the latest SP will fix some of the issues. But in reality, this is truly a design issue -- one that will require somewhat of a redesign on how CFMX services manage themselves, report/handle errors and runaway conditions, etc. As we all know, the Restart Server on X Unresponsive Requests simply has never worked. I think the current behaviour is OK, as long as the Restart Server on X Unresponsive Requests works. I never have that many hung threads so I can't confirm or deny. Jochem ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
RE: CFMX Server Crash #4 for the day..
Absolutely. That's something we tested, and it did seem to make a slight difference. It did not stop things like cyclical calls to a page -- which, in reality, do happen and should be managed by the server... rather than simply being brought to it's knees. | Piling up some assumptions: | | First of all, CF MX will wait indefinitely for database queries to | return if there is a problem in the JDBC driver or the database that | does not return an error condition. This will add to the | running request | number as well as eat 1 client access licence from the | dataabse server. This means that following request will have | less threads to work with, | and have to open a new databse connection. Eventually, this database | connection also breaks. Next client access license + thread eaten. | Process repeats itself until the maximum number of allowed | requests is | reached. | | Some way to test this scenario would be to disable connection | pooling. | Then you should be able to see on the database side how long | connections | stay open. If they stay open past the longest query, you found the | problem. | Another option would be to try and limit the amount of connections to | the database to something less as the restart threshold. Then | you would | start to see timeouts from queries not being able to get a database | connection before the server actually crashes. That's just the point, it doesn't work. And to my knowledge it has *never* worked in any version of CF. Certainly not that I could ever see. The server could be brought down thru DOS attacks. And essentially, this is what is happening -- albeit not on purpose. | At any rate.. The theory is that the latest SP will fix some of the | issues. But in reality, this is truly a design issue -- | one that will | require somewhat of a redesign on how CFMX services manage | themselves, | report/handle errors and runaway conditions, etc. As we | all know, the | Restart Server on X Unresponsive Requests simply has never worked. | | I think the current behaviour is OK, as long as the Restart | Server on X | Unresponsive Requests works. I never have that many hung | threads so I | can't confirm or deny. | ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq This list and all House of Fusion resources hosted by CFHosting.com. The place for dependable ColdFusion Hosting.
RE: CFMX Server Crash #4 for the day..
ugh, never worked? I beg to differ. I have not moved to CFMX yet, but this setting works. I have an app that is very DB intense and I have watched the logs pile up with unreposnsive thread count has reached x and when x = the restart server on y unresponsive requests the CF process did restart. In my experience this was DB relatedlong queries, Db down, etc. After working on the app we did the following: 1) worked on better SQL 2) raised the request time out limit per request and server wide 3) tweakd the Restart Server on X Unresponsive Requests setting in combo with the number of request limit. Now we occasionaly get an entry in the CF logs with unreposnsive thread count has reached x where x is 0 or 1 and no server hangs. I don't mean to be crass, but you got to do some work on this stuff, it may not be CF's fault. Doug Restart Server on X Unresponsive Requests simply has never worked. I think the current behaviour is OK, as long as the Restart Server on X Unresponsive Requests works. I never have that many hung threads so I can't confirm or deny. Jochem ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Get the mailserver that powers this list at http://www.coolfusion.com
RE: CFMX Server Crash #4 for the day..
Doug, you are being Crass, and worse yet your are being so without understanding the scope of this conversation which makes your being crass unjustified. I am re-posting an earlier post from Lee (who was the one who made the restart after X never did work comment. And you tell me if this qualifies as an example of some one working towards a solution or just blaming CF. ~| Archives: http://www.houseoffusion.com/cf_lists/index.cfm?forumid=4 Subscription: http://www.houseoffusion.com/cf_lists/index.cfm?method=subscribeforumid=4 FAQ: http://www.thenetprofits.co.uk/coldfusion/faq Signup for the Fusion Authority news alert and keep up with the latest news in ColdFusion and related topics. http://www.fusionauthority.com/signup.cfm
RE: CFMX Server Crash #4 for the day..
The body of the message was cut off, here it is again: date: Fri, 25 Oct 2002 14:58:52 -0700 from: Lee Fuller [EMAIL PROTECTED] Subject: CFMX Issue(s) Status Reply-To: [EMAIL PROTECTED] To: CF-Talk [EMAIL PROTECTED] Hello everyone.. Thought I would update you on the issues that caused the long thread here. I'm posting here, because this is NOT about how to manage communication between MM and ourselves. Rather, this is specifically about the technical aspects of what's happening now. Here goes... (This is a long one!) ** BIG - HUGE - HONKIN' - DISCLAIMER I'm sending out this info for just that purpose.. INFO. It is NOT recommended as a way for you to handle your own problems with CF(MX), nor is this information, in any way shape or form, given at the behest of, or suggestion from, Macromedia. These are MY observations and workarounds. THEY MAY NOT WORK IN YOUR CASE -- PARTIALLY OR AT ALL, AND MAY DESTROY DATA, GIVE YOU A HORRIBLE, PAINFUL AND INCURABLE DISEASE, REMOTELY KILL YOUR PARENTS, GRANDPARENTS AND/OR REMAINING FAMILY WITH NUCLEAR FALLOUT, AND/OR DRAIN YOUR BANK ACCOUNT, ALONG WITH ANY NUMBER OF OTHER MORE PAINFUL AFTER-EFFECTS ((c)2002 ADOBE). You have been warned. * Issue #1: ~ Server Spiraling Out Of Control Symptom: At irregular intervals, IIS5 will begin to spiral upwards in the number of Current Connections, showing that many of them are not dead/dying properly. Correspondingly, the number of Running Requests of the JRUN (CFMX) service begin to rise as well, with a floor number staying stable (not dying), yet ever increasing until the CFMX server will stop responding to CFM file requests at all. If we restart only IIS, on some occassions CFM files CALLED DIRECTLY will begin to fail with 404 errors. However, indirect calls to the same files will work normally (sometimes). I.e., http://www.somedomain.com/index.cfm; fails with a 404, whereas http://www.somedomain.com; works fine (sometimes) (and yes.. there is ONLY an index.cfm file -- no default.xxx or other home pages in the dir). Workaround: ~~~ We have noticed that simply using the PerfMonitor to restart the CFMX and IIS services does not always work properly. Sometimes, as mentioned above, the IIS server simply loses touch with CFMX and delivers bizarre 404 errors, where other times CFMX will simply not respond and the entire site appears to be offline. This DOES NOT correspond (as far as we can tell) in any significant way to whether the page connecting to an outside datasource. Therefore, we decided to try simply restarting the jrun.exe service directly using the cfstop.bat and cfstart.bat files that are located in the CFMX Root Dir\bin directory. We trigger this restart using PerfMon and setting alerts to make sure that we have no more than 25 Running Requests for the JRUN service, and another that runs the restart if we have more than 125 Current Connections to IIS. These numbers may seem low. However, if we do not restart at these points, either IIS or CFMX will begin to run VERY slowly, or simply not respond at all. And since CFMX's Probes stop working when CFMX itself stops working, a probe is useless. Besides, CF never does restart itself properly ... after X unresponsive requests. Here is the BAT file that is used to restart the JRUN service: === @echo off cls REM == RESTART CFMX === c: cd \ROOTCFMXDIR\bin call C:\ROOTCFMXDIR\bin\cfstop.bat c:\winnt\system32\kill -f jrun.exe c:\winnt\system32\kill -f jrun.exe sleep 2 call C:\ROOTCFMXDIR\bin\cfstart.bat REM === Exit (interesting that this isn't needed, nor does it work.. So don't ask me why it's here. :) === You may well ask Why do you kill JRUN.EXE twice? Good question! Because JRUN.EXE is an actual service, we are effectively killing the old service and opening a new one in a new CMD window (hidden of course). We noticed that if we only kill it one time, it will not die. Therefore, we will end up with dozens of CMD instances, along with dozens of JRUN.EXE instances by day's end. We have found that killing it twice does, indeed, get rid of the old instance, and properly create only one new instance. It is important to note that there is no way (at least that I know of) to cause PerfMon alerts to start themselves, upon a reboot caused remotely. So if you're not there to start the alerts manually, they will not restart the server if they are triggered. Lastly, we tried adding the CFMX ODBC services to this restart bat file.. with no consequence other than it taking longer to come back to life. So we took them back out. Currently, this work around causes the server to go offline for about 10-15 seconds... each time it is restarted. Status with Macromedia: ~~~ Pending solution. - Issue #2: ~ NTEXT Fields in SQL Database Not Being Returned