Re: Strange intermittent WebDNA problems
This WebDNA talk-list message is from 2008
It keeps the original formatting.
numero = 70082
interpreted = N
texte = Palle Bo Nielsen wrote: > Are you saying that you have found a method which makes WebDNA > compatible with Leopard - and furthermore maybe on an Intel Mac? > > Palle leopard on G5.. O.K., I'll start, 6.0a, unix-Linux Intel RedHat (on Fedora) About 2 weeks ago to this day, I started having issues with a clients WebSite where the CPU usage would hop up to 100% usage, the http processes would also build up. This started soon after (about an hour) a data/time stamp was munged on the OS server (it was speculated that it synced to a time server). I am not the admin of the server... just the WebDNA guy, so I don't have all the details there. It was also speculated that this time-stamp messed with the safe-write system in WebDNA. So, for the databases that were utilizing the safe-write system (you can tell if they are utilizing the safe write system by the build up of -tmp suffixed databases in the same directory as your regular database). One database was entirely gone and we replaced by using an existing -tmp file that the "safe-write" system had written. I turned off "safe-write" at that time. From there, we were still having problems. I chased after seg-fault issues (bad code) that I found using the file created from the signal catcher in the pref. (it appears that many builds are not as verbose as the one I am using (listed above)). I can't really use the "log error messages" preference because the site I am dealing with uses the errorlogmessages.db to create dynamic pages from calls that are not found... (so turning it on creates a huge file of mostly acceptable stuff really quickly). I also did some strace's when I noticed the httpd processes jump up. In the end, these helped keep the site up longer, but did not solve the problem. I then started taking a look at the databases. Since I had turned off "safe-write" the database issues made themselves better known. I started getting some truncated databases. I looked further. There were a lot of control characters and such... zapping gremlins and cleaning out the databases again seemed to help performance. However, these are greater than 30MB databases and there is still some remaining odd characters because I left the encrypted passwords in there. I was still getting truncated databases but not as often. At this point, I then took a look at how these passwords were being encrypted and found they they were not being saved to the database in the correct way that we (on the list) have come to know. So I am currently in the process of having everyone reset their passwords and saving them correctly! I turned back on the safe-write system as well... because we needed to give our customers a respite from all the problems! This is a new site I acquired so I am just getting used to it. Lots of database interactions! One thing that we discovered is that upon reboot, the WebDNAMonitor was not starting (it shows up as another "webcatalog" in our process readings). Once we realized this and manually start it each time we reboot, the WebServer amazingly stays up and the http builds ups do not happen! It appears to me that the Safe-Write system and the WebDNAMonitor act as big brother and hide a lot of little mistakes that are harmful to a healthy operating WebDNA site. I am not stopping now with healing this site just because it appears to be working. I am going to install exclusivelocks on just about every crucial database interaction ([appends..],[deletes..] [replaces..] etc..) I am also going to cleanse the existing data (part of the password resetting process). So you see, there is a lot that can go into stalling a server! Other things to consider are large files that are being interacted with regarding apache. (I think there is a 2gig default limit on apache served files). Your turn! Donovan P.S. wrote this really fast!.. so sorry about spelling. -- =o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o DONOVAN D. BROOKE EUCA Design Center <- Web Development (specializing in eCommerce),-> <- Desktop Publishing, Print Consulting, Labels -> Web:> http://www.egg.bz | http://www.euca.us =o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o ------------------------------------------------------------- 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://webdna.smithmicro.com/
Associated Messages, from the most recent to the oldest:
Palle Bo Nielsen wrote: > Are you saying that you have found a method which makes WebDNA > compatible with Leopard - and furthermore maybe on an Intel Mac? > > Palle leopard on G5.. O.K., I'll start, 6.0a, unix-Linux Intel RedHat (on Fedora) About 2 weeks ago to this day, I started having issues with a clients WebSite where the CPU usage would hop up to 100% usage, the http processes would also build up. This started soon after (about an hour) a data/time stamp was munged on the OS server (it was speculated that it synced to a time server). I am not the admin of the server... just the WebDNA guy, so I don't have all the details there. It was also speculated that this time-stamp messed with the safe-write system in WebDNA. So, for the databases that were utilizing the safe-write system (you can tell if they are utilizing the safe write system by the build up of -tmp suffixed databases in the same directory as your regular database). One database was entirely gone and we replaced by using an existing -tmp file that the "safe-write" system had written. I turned off "safe-write" at that time. From there, we were still having problems. I chased after seg-fault issues (bad code) that I found using the file created from the signal catcher in the pref. (it appears that many builds are not as verbose as the one I am using (listed above)). I can't really use the "log error messages" preference because the site I am dealing with uses the errorlogmessages.db to create dynamic pages from calls that are not found... (so turning it on creates a huge file of mostly acceptable stuff really quickly). I also did some strace's when I noticed the httpd processes jump up. In the end, these helped keep the site up longer, but did not solve the problem. I then started taking a look at the databases. Since I had turned off "safe-write" the database issues made themselves better known. I started getting some truncated databases. I looked further. There were a lot of control characters and such... zapping gremlins and cleaning out the databases again seemed to help performance. However, these are greater than 30MB databases and there is still some remaining odd characters because I left the encrypted passwords in there. I was still getting truncated databases but not as often. At this point, I then took a look at how these passwords were being encrypted and found they they were not being saved to the database in the correct way that we (on the list) have come to know. So I am currently in the process of having everyone reset their passwords and saving them correctly! I turned back on the safe-write system as well... because we needed to give our customers a respite from all the problems! This is a new site I acquired so I am just getting used to it. Lots of database interactions! One thing that we discovered is that upon reboot, the WebDNAMonitor was not starting (it shows up as another "webcatalog" in our process readings). Once we realized this and manually start it each time we reboot, the WebServer amazingly stays up and the http builds ups do not happen! It appears to me that the Safe-Write system and the WebDNAMonitor act as big brother and hide a lot of little mistakes that are harmful to a healthy operating WebDNA site. I am not stopping now with healing this site just because it appears to be working. I am going to install exclusivelocks on just about every crucial database interaction ([appends..],[deletes..] [replaces..] etc..) I am also going to cleanse the existing data (part of the password resetting process). So you see, there is a lot that can go into stalling a server! Other things to consider are large files that are being interacted with regarding apache. (I think there is a 2gig default limit on apache served files). Your turn! Donovan P.S. wrote this really fast!.. so sorry about spelling. -- =o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o DONOVAN D. BROOKE EUCA Design Center <- Web Development (specializing in eCommerce),-> <- Desktop Publishing, Print Consulting, Labels -> Web:> http://www.egg.bz | http://www.euca.us =o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o=o ------------------------------------------------------------- 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://webdna.smithmicro.com/
Donovan Brooke
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:
Dynamic Generation of PDFs? (2003)
Plugin or CGI or both (1997)
What is the best way to share load with Web Cat and mac (2000)
Stopping bad HTML propagation ? (1997)
E-mail loop ! (1997)
Multipart/alternative e-mail (2003)
Working HTML email (2001)
Summing a field full of numbers ... (1997)
Protect (1997)
2.0 Info (1997)
OT: Google image search results (2004)
Hosting Services (1999)
WebCat2 Append problem (B14Macacgi) (1997)
2.1b3 --> way slow (1997)
Maybe minor bug report (1997)
WebCatalog2 Feature Feedback (1996)
PGP encrypted email (1998)
WebCat2b13MacPlugin - nested [xxx] contexts (1997)
PIXO (1997)
syntax question, not in online refernce (1997)