Re: [WebDNA] v7 Stress Test and Strange Behaviour
This WebDNA talk-list message is from 2012
It keeps the original formatting.
numero = 107969
interpreted = N
texte = > What Aaron describe is that he populates a speed.db with a specific =content (let's name it "A"), read it and it shows well. Then he =populates the same speed.db with a "B" content, reads it, and it shows =well. Then he populates the same speed.db with a "C" content, reads it, =and it shows well.>=20> Afterwards he lists the speed.db to show its content and randomly gets =the content "A", "B" or "C"Correct. There is a "show only" page which performs no other actions =but [search][founditems] of speed.db. Each page does show its own =records correctly and consistently. But if I load each of the pages =consecutively, and come back to "Show", then I'll randomly get content =from "A" "B" or "C". =20A few more notes and observations: =20-Sometimes, it'll take a dozen or so page loads before the strange =behavior appears with the stress test. Sometimes, it only takes one or =two. Sometimes, it'll seem to be working perfectly, but as I've =mentioned before, it always eventually breaks.-As I've mentioned before, I even see this with the webdna.ini db. If I =make a change on the adminprefs, and reload the page a couple times, =I'll even see my prefs "bounce around" upon consecutive refreshes-I found that the actual file on disk is changing too. If I append a =few records to a db, then check the file, it appears correct. I can =refresh my browser and it will list the wrong records, but the file is =correct. Sometimes its the other way around. And then sometimes, I'll =open the file and will be correct and then open again and it has =reverted back to an older version. =20-If I walk away from this for say a half hour or something and come back =to it, it'll appear to have corrected itself. Once I start putting it =to the tests though, it always eventually fails. =20-The flushdatabases command seems to have no affect here. I've flushed =dbs and templates, cleared my cache, restarted apache, flushed again, =and came back to my "Show" page, and would see the errors with page =refreshes without even modifying the db first. =20>=20> No one else has been able to reproduce it, because it seems that if =you use a remote browser, the strange behaviour does not occur: only =locally. I did test it locally too, but have had no problem.Its a VPS, so I am only accessing remotely. I've reproduced the problem =on every computer that I have around. For some reason, the stress test =(with speed.db) will actually pass on several attempts. But modifying =adminprefs (as well as some other dbs), I see corruption almost =immediately. Perhaps no else has seen this because they are not testing =as rigorously as I am. If the test passes 20 times, I am still =skeptical, and will test 20 more times. It always eventually fails.> Aaron also noticed the behaviour disappears with Ubuntu 11 but is =present with Ubuntu 10 and Debian 6.This Debian 6 VPS is LAMP configured (pre-configured by DreamHost) and =there are a lot of sites running plus PHP fcgi, and so on. There =certainly could be something on there I don't know about that's causing =a problem. However, this behavior is exactly the same that I experienced =in Ubuntu 10. In that previous case, it was a completely clean install, =plain-vanilla config. I tried the exact same thing with Ubuntu 11, and =did not get this behavior. The only thing I observed that was different =between all of these was the libssl.so version. Both Ubuntu 10 and =Debian 6 came with libssl.so.0.9.8. On Ubuntu 11 though, it is =libssl.so.1.0.0, and I had to download libssl.so.0.9.8 to get WebDNA7 =working. I don't know if that has any relevance or not.Also, could the fact that these are VPS make a difference? That is, as =opposed to administering the physical machine, which it seems some of =you (who use v7) are doing. =20Lastly, I understand that there are two different fast cgi modules. The =one that I've been using is fcgid, but is there a WebDNA7 configuration =for the other (fastcgi or fcgi if I recall)? I think it is something to =try. =20I don't think its going to be possible to change to lighttpd on this =VPS. There are already 35 other sites running and some of them use =WordPress, Joomla, etc and are highly dependent on mod_rewrite and =.htaccess. =20Chris, I'm sending you more info off list.Thanks everyone!On Jan 7, 2012, at 10:08 PM, Kenneth Grome wrote:>> No one else has been able to reproduce it, because it>> seems that if you use a remote browser, the strange>> behaviour does not occur: only locally.>=20> How was he able to use a local browser on a VPS?>=20> Sincerely,> Kenneth Grome> ---------------------------------------------------------> This message is sent to you because you are subscribed to> the mailing list
.> To unsubscribe, E-mail to: > archives: http://mail.webdna.us/list/talk@webdna.us> Bug Reporting: support@webdna.us
Associated Messages, from the most recent to the oldest:
> What Aaron describe is that he populates a speed.db with a specific =content (let's name it "A"), read it and it shows well. Then he =populates the same speed.db with a "B" content, reads it, and it shows =well. Then he populates the same speed.db with a "C" content, reads it, =and it shows well.>=20> Afterwards he lists the speed.db to show its content and randomly gets =the content "A", "B" or "C"Correct. There is a "show only" page which performs no other actions =but [search][founditems] of speed.db. Each page does show its own =records correctly and consistently. But if I load each of the pages =consecutively, and come back to "Show", then I'll randomly get content =from "A" "B" or "C". =20A few more notes and observations: =20-Sometimes, it'll take a dozen or so page loads before the strange =behavior appears with the stress test. Sometimes, it only takes one or =two. Sometimes, it'll seem to be working perfectly, but as I've =mentioned before, it always eventually breaks.-As I've mentioned before, I even see this with the webdna.ini db. If I =make a change on the adminprefs, and reload the page a couple times, =I'll even see my prefs "bounce around" upon consecutive refreshes-I found that the actual file on disk is changing too. If I append a =few records to a db, then check the file, it appears correct. I can =refresh my browser and it will list the wrong records, but the file is =correct. Sometimes its the other way around. And then sometimes, I'll =open the file and will be correct and then open again and it has =reverted back to an older version. =20-If I walk away from this for say a half hour or something and come back =to it, it'll appear to have corrected itself. Once I start putting it =to the tests though, it always eventually fails. =20-The flushdatabases command seems to have no affect here. I've flushed =dbs and templates, cleared my cache, restarted apache, flushed again, =and came back to my "Show" page, and would see the errors with page =refreshes without even modifying the db first. =20>=20> No one else has been able to reproduce it, because it seems that if =you use a remote browser, the strange behaviour does not occur: only =locally. I did test it locally too, but have had no problem.Its a VPS, so I am only accessing remotely. I've reproduced the problem =on every computer that I have around. For some reason, the stress test =(with speed.db) will actually pass on several attempts. But modifying =adminprefs (as well as some other dbs), I see corruption almost =immediately. Perhaps no else has seen this because they are not testing =as rigorously as I am. If the test passes 20 times, I am still =skeptical, and will test 20 more times. It always eventually fails.> Aaron also noticed the behaviour disappears with Ubuntu 11 but is =present with Ubuntu 10 and Debian 6.This Debian 6 VPS is LAMP configured (pre-configured by DreamHost) and =there are a lot of sites running plus PHP fcgi, and so on. There =certainly could be something on there I don't know about that's causing =a problem. However, this behavior is exactly the same that I experienced =in Ubuntu 10. In that previous case, it was a completely clean install, =plain-vanilla config. I tried the exact same thing with Ubuntu 11, and =did not get this behavior. The only thing I observed that was different =between all of these was the libssl.so version. Both Ubuntu 10 and =Debian 6 came with libssl.so.0.9.8. On Ubuntu 11 though, it is =libssl.so.1.0.0, and I had to download libssl.so.0.9.8 to get WebDNA7 =working. I don't know if that has any relevance or not.Also, could the fact that these are VPS make a difference? That is, as =opposed to administering the physical machine, which it seems some of =you (who use v7) are doing. =20Lastly, I understand that there are two different fast cgi modules. The =one that I've been using is fcgid, but is there a WebDNA7 configuration =for the other (fastcgi or fcgi if I recall)? I think it is something to =try. =20I don't think its going to be possible to change to lighttpd on this =VPS. There are already 35 other sites running and some of them use =WordPress, Joomla, etc and are highly dependent on mod_rewrite and =.htaccess. =20Chris, I'm sending you more info off list.Thanks everyone!On Jan 7, 2012, at 10:08 PM, Kenneth Grome wrote:>> No one else has been able to reproduce it, because it>> seems that if you use a remote browser, the strange>> behaviour does not occur: only locally.>=20> How was he able to use a local browser on a VPS?>=20> Sincerely,> Kenneth Grome> ---------------------------------------------------------> This message is sent to you because you are subscribed to> the mailing list .> To unsubscribe, E-mail to: > archives: http://mail.webdna.us/list/talk@webdna.us> Bug Reporting: support@webdna.us
aaronmichaelmusic@gmail.com
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:
My first ship table (1998)
test (2000)
Giving out error pages (1997)
RE: Purchased cart being overwritten (1997)
Problems appending to database (1997)
Contest (1999)
Mac OSX 10.5 with WebSTAR... Webcatalog won't run (2003)
Country & Ship-to address & other fields ? (1997)
[foundItems] context problem (2000)
I'm tired of all this! (2000)
[WebDNA] Sendmail 'return-path' header that is different to the 'from' header (2009)
shipcost (1997)
how do I delete 1 of 2 identical records? (2003)
TCPConnect (2002)
Help name our technology! (1997)
incrementing a counter remotely? (1999)
Enhancement Request for WebCatalog-NT (1996)
I think WebDNA-Talk died ... :( (1997)
6.1 Patch - Form Data Back Button Issue (2006)
Thanks Grant (1997)