Re: Stability complaints

This WebDNA talk-list message is from

2002


It keeps the original formatting.
numero = 40411
interpreted = N
texte = Dale,Thank you for sharing your detailed analysis. I'm pushing about 7-10 GB of traffic a month right now and I get some crashes, but it's not been a daily thing - however, any crash is not really acceptable.I have two questions:1. Where is the error log that you read the output from the crash? 2. Will you share the WebCatalog restart script? I know the command to restart the server, but how do you run the script on crash?Running : OS 10.1.2, Webcat 4.0.something- Ben Swihart Gotricities Network-----Original Message----- From: WebCatalog Talk [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf Of Dale LaFountain Sent: Monday, June 03, 2002 5:11 PM To: WebCatalog Talk Subject: Re: Stability complaints >Bob Minor wrote: >>BTW my webcat/windows applications run fine, but when running on MacOSX, >>webcat dies every few days, so go figure. >> > >WebCat runs under a variety of O/S's, each with their own >idiosyncracies. If you can produce code which works on one O/S and >not on another, I'm sure that SMSI would be very happy to fix. If >it is not reproducable, it didn't happen (you can call that >Peacock's Law if you like ;~). The vast majority of the stability >discussion here is WebCat's not stable for me on ___ with little >or no details that could be used to reproduce the problems and fix >it. Even the most basic triage seems to be impossible to pry out of >people on this list: > > 1) is the web server still serving static pages? > 2) what version of ___ are you running (c library, O/S, web server, etc)? > 3) are you logging any suspicious activity (CodeRed worm, etc)? > 4) have you confirmed you have free disk space, lots of RAM, no >thermal problems, any other hardware issues? > 5) have you turned on the extended WC logging to track which pages >are failing? > > >>BTW are you running any webcat templates under *nix, so that you can make an >>experienced assesment, or are you just talking in general about webcat? >> > >My production server is NT (not even W2k), so I am speaking more >generally about WC. I have run WC in test mode under Linux, but I >am happy with my NT web server (WebSite) and am not ready to switch. >I also run almost all of my large pure text files and graphics out >of a Linux box that gets rebooted when the power in the building >goes out (too frequently for my taste). I _will_ move to Linux at >some point. > >JohnI have been testing in OSX since Jan 2002 under various conditions, and running in production for a couple months.The only thing I can say for sure is that a dual 1Ghz crashes far less often than a G4/500 running the same code under the same traffic loads. Something about concurrent requests not happening nearly as often with the dual gig in place leads me to suspect webcat's ability to deal with multiple simultaneous requests under heavy load...I have one OSX box that is crashing multiple times per day, each time on a different page than the last according to webcat.debug (sound familiar anyone?). This code ran on an OS9/Webcat402 box for weeks at a time without crashing.As for debugging, I have tried multiple versions of the OS from 10.1.1 through 10.1.4, webcat versions from 4.0.2 through the latest test builds from Scott, I have swapped out the hardware entirely, tested on specific templates with Netsplat for innumerable hours, and had several experienced developers comb through the site's code, including the original authors. Swapping hardware did reduce the crashes somewhat, but ANY number of crashes that happen every day are too many.This is an OSX-specific problem from what I hear from SMSI, specifically with memory allocation. Based on all the other successful (non-crashing) apps on OSX, one would think that whatever os-specific issues could be worked around until they are properly fixed by Apple or the open source developers. Speaking of crashing, I just crashed again (#6 today). Here's the output from webcat dying:*** malloc[18292]: error for object 0x45af770: Incorrect check sum for freed object - object was probably modified after beeing freed; break at szone_error sem_wait failed: returned EINVAL: Invalid argument sem_wait failed: returned EINVAL: Invalid argument *** malloc[18292]: Deallocation of a pointer not malloced: 0xf2c90; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debugLooked a little further into the above, and it looks like I was being spidered. Blocked the spider, and now the crashes have stopped. Valid pages were being requested, just at too high of a rate for webcat to deal with nicely. I'm guessing another dual 1ghz g4 would solve some or most of this problem, but I don't have the cash at the moment...My restart script gets webcat back up and running within 60 seconds max, so the downtime is not horrible. Nevertheless, it still sucks that I'm crashing this much under OSX. I'm grateful for all the work that SMSI has put into improving WebCat OSX so far, but from what I can see webcat still needs some help in the stability department.-Dale------------------------------------------------------------- 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:

    
  1. Re: Stability complaints (Bob Minor 2002)
  2. Re: Stability complaints (Dale LaFountain 2002)
  3. Re: Stability complaints (Bob Minor 2002)
  4. Re: Stability complaints (Ben Swihart 2002)
  5. Re: Stability complaints (Dale LaFountain 2002)
Dale,Thank you for sharing your detailed analysis. I'm pushing about 7-10 GB of traffic a month right now and I get some crashes, but it's not been a daily thing - however, any crash is not really acceptable.I have two questions:1. Where is the error log that you read the output from the crash? 2. Will you share the WebCatalog restart script? I know the command to restart the server, but how do you run the script on crash?Running : OS 10.1.2, Webcat 4.0.something- Ben Swihart Gotricities Network-----Original Message----- From: WebCatalog Talk [mailto:WebDNA-Talk@talk.smithmicro.com]On Behalf Of Dale LaFountain Sent: Monday, June 03, 2002 5:11 PM To: WebCatalog Talk Subject: Re: Stability complaints >Bob Minor wrote: >>BTW my webcat/windows applications run fine, but when running on MacOSX, >>webcat dies every few days, so go figure. >> > >WebCat runs under a variety of O/S's, each with their own >idiosyncracies. If you can produce code which works on one O/S and >not on another, I'm sure that SMSI would be very happy to fix. If >it is not reproducable, it didn't happen (you can call that >Peacock's Law if you like ;~). The vast majority of the stability >discussion here is WebCat's not stable for me on ___ with little >or no details that could be used to reproduce the problems and fix >it. Even the most basic triage seems to be impossible to pry out of >people on this list: > > 1) is the web server still serving static pages? > 2) what version of ___ are you running (c library, O/S, web server, etc)? > 3) are you logging any suspicious activity (CodeRed worm, etc)? > 4) have you confirmed you have free disk space, lots of RAM, no >thermal problems, any other hardware issues? > 5) have you turned on the extended WC logging to track which pages >are failing? > > >>BTW are you running any webcat templates under *nix, so that you can make an >>experienced assesment, or are you just talking in general about webcat? >> > >My production server is NT (not even W2k), so I am speaking more >generally about WC. I have run WC in test mode under Linux, but I >am happy with my NT web server (WebSite) and am not ready to switch. >I also run almost all of my large pure text files and graphics out >of a Linux box that gets rebooted when the power in the building >goes out (too frequently for my taste). I _will_ move to Linux at >some point. > >JohnI have been testing in OSX since Jan 2002 under various conditions, and running in production for a couple months.The only thing I can say for sure is that a dual 1Ghz crashes far less often than a G4/500 running the same code under the same traffic loads. Something about concurrent requests not happening nearly as often with the dual gig in place leads me to suspect webcat's ability to deal with multiple simultaneous requests under heavy load...I have one OSX box that is crashing multiple times per day, each time on a different page than the last according to webcat.debug (sound familiar anyone?). This code ran on an OS9/Webcat402 box for weeks at a time without crashing.As for debugging, I have tried multiple versions of the OS from 10.1.1 through 10.1.4, webcat versions from 4.0.2 through the latest test builds from Scott, I have swapped out the hardware entirely, tested on specific templates with Netsplat for innumerable hours, and had several experienced developers comb through the site's code, including the original authors. Swapping hardware did reduce the crashes somewhat, but ANY number of crashes that happen every day are too many.This is an OSX-specific problem from what I hear from SMSI, specifically with memory allocation. Based on all the other successful (non-crashing) apps on OSX, one would think that whatever os-specific issues could be worked around until they are properly fixed by Apple or the open source developers. Speaking of crashing, I just crashed again (#6 today). Here's the output from webcat dying:*** malloc[18292]: error for object 0x45af770: Incorrect check sum for freed object - object was probably modified after beeing freed; break at szone_error sem_wait failed: returned EINVAL: Invalid argument sem_wait failed: returned EINVAL: Invalid argument *** malloc[18292]: Deallocation of a pointer not malloced: 0xf2c90; This could be a double free(), or free() called with the middle of an allocated block; Try setting environment variable MallocHelp to see tools to help debugLooked a little further into the above, and it looks like I was being spidered. Blocked the spider, and now the crashes have stopped. Valid pages were being requested, just at too high of a rate for webcat to deal with nicely. I'm guessing another dual 1ghz g4 would solve some or most of this problem, but I don't have the cash at the moment...My restart script gets webcat back up and running within 60 seconds max, so the downtime is not horrible. Nevertheless, it still sucks that I'm crashing this much under OSX. I'm grateful for all the work that SMSI has put into improving WebCat OSX so far, but from what I can see webcat still needs some help in the stability department.-Dale------------------------------------------------------------- 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/ Ben Swihart

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:

Another webcatalog site!! (1997) Error Msg (1998) form data submission gets truncated (1997) Newbie questions (1999) Emailer help....! (1997) ConvertChars (2000) Add - optional parameters textA=.... (1997) unsubscribe (1997) MacAuthorize order data fields WAS:How To question... (1997) emailer (1997) [WebDNA] tag [validcard] fails on webdna 7.0. Do I need the commerce edition? (2011) Search in 2 or more catalogs (1997) 2.1b3 --> way slow (1997) Trigger Happy (2005) RE: WebCatalog2 for NT Beta Request (1997) no global [username] or [password] displayed ... (1997) Vanity counter side note (1998) restarting service remotely on NT (1997) This Code Kills My WebCatalog Dead (2003) [WebDNA] calling a webDNA tmpl from PHP? (2008)