Re: [WebDNA] sem_open bugfix for os x version 6.2
This WebDNA talk-list message is from 2012
It keeps the original formatting.
numero = 108240
interpreted = N
texte = Rob, just curiosity: what formatting do you use for your drive? = (thinking about Returned ENOSPC) About the semaphore bug, here a little more explanations: It was due to a poor method of generating the semaphore names. The = semaphore name was generated this way: sprintf(mSemName, "webcat_%ld_%ld", this, times(&start)); The name therefore is "webcat_" + address space location of current = object + current time in seconds. The issue is that on non-ASLR platforms the address spaces are identical = each time the program runs. And since Apache starts all of the FastCGI = processes at the same time, the times are also often identical. So basically WebDNA is trying to open semaphores at the same times with = the same names. When it doesn't work, it starts spewing out all kinds of = errors. After it fails to open the semaphore, it keeps trying to use it = anyway and try over and over to use a semaphore it never opened - chris On Jan 25, 2012, at 14:13, Rob wrote: > Donovan, >=20 > There are two of us, that I know of, that are having all kinds of = weird issues that a lot of times don't even relate to any website, and = are willing to pay Chris for a solution as it's consuming all are time. = For example, I can go to one of my sites home page one time, and it = crashes WebCatalog, then the next time I go its fine. The next time I go = to the Preferences in the admin, and its fine, if I go again it crashes. = This is in the middle of the night with low to no traffic. >=20 > I've reinstalled about three times, I've gone through all the sites = and made sure there was no issues, and checked all the permissions on = the db's and files etc. All of these sites run perfectly fine on my = development box, which is actually Lion, and I don't get the issue in = terminal, like I do in Snow Leopard. If there was one file I could = replace, or even remove in Apache or something, I would do that, but = there is nothing I can put my finger on. Whatever it is is related to = whats happening in terminal, which shows the following when you restart = WebDNA: > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device >=20 > Rob >=20 >=20 > On Jan 25, 2012, at 9:41 AM, Donovan Brooke wrote: >=20 >> First, I wouldn't call something a bug unless I can confirm and = reproduce it. Why exactly are "we" calling this a bug in 6.2? How is it = reproduced? >>=20 >> To answer below: >>=20 >> [snip] >>=20 >> Also, we don't know if there are other revisions or fixes Donovan = team may have done for 6.2 that aren't in our SVN. >>>>=20 >>>> I suggest you speak privately (or publicly) with Rob Forest who has = the same problem with his 6.2 copy: a group of several users could club = together to get it fixed. Our charge would be somewhere between $800 to = $1200 >>=20 >>=20 >> I think Chris knows that I'd be interested in a build/patch to = CICADA.. not for the stated "bug" really, but for the Apache2 fixes that = we made to the plugin (content type text/XML), and the SMTP auth fixes. = Those should both already be done in the 6x branch and awaiting a build. >>=20 >> There may be some other fixes that were made, and there may be some = loose end's as well... since there was some "changes" going on with WSC = policy about the time 6x was tagged and when I left. I don't know = exactly what state 6.2 was tagged at, but it should be doable without = too much problem. The developer would want to make sure they use the = apache module with the fix on the content type. >>=20 >> We were also playing around with process isolation.. much like how = WebDNA7 works. WebDNA 6.2 does not have to be installed server-wide and = we did have a build working that would allow multiple installs on a = single server. Not sure if this was kept or not. >>=20 >> Anyway... ahh, memories. :-) >>=20 >> Donovan >>=20 >>=20 >> --=20 >> Donovan Brooke >> Euca Design Center >> www.euca.us >>=20 >> --------------------------------------------------------- >> 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 >=20 > --------------------------------------------------------- > 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:
Rob, just curiosity: what formatting do you use for your drive? = (thinking about Returned ENOSPC) About the semaphore bug, here a little more explanations: It was due to a poor method of generating the semaphore names. The = semaphore name was generated this way: sprintf(mSemName, "webcat_%ld_%ld", this, times(&start)); The name therefore is "webcat_" + address space location of current = object + current time in seconds. The issue is that on non-ASLR platforms the address spaces are identical = each time the program runs. And since Apache starts all of the FastCGI = processes at the same time, the times are also often identical. So basically WebDNA is trying to open semaphores at the same times with = the same names. When it doesn't work, it starts spewing out all kinds of = errors. After it fails to open the semaphore, it keeps trying to use it = anyway and try over and over to use a semaphore it never opened - chris On Jan 25, 2012, at 14:13, Rob wrote: > Donovan, >=20 > There are two of us, that I know of, that are having all kinds of = weird issues that a lot of times don't even relate to any website, and = are willing to pay Chris for a solution as it's consuming all are time. = For example, I can go to one of my sites home page one time, and it = crashes WebCatalog, then the next time I go its fine. The next time I go = to the Preferences in the admin, and its fine, if I go again it crashes. = This is in the middle of the night with low to no traffic. >=20 > I've reinstalled about three times, I've gone through all the sites = and made sure there was no issues, and checked all the permissions on = the db's and files etc. All of these sites run perfectly fine on my = development box, which is actually Lion, and I don't get the issue in = terminal, like I do in Snow Leopard. If there was one file I could = replace, or even remove in Apache or something, I would do that, but = there is nothing I can put my finger on. Whatever it is is related to = whats happening in terminal, which shows the following when you restart = WebDNA: > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device > sem_open failed: returned ENOSPC: No space left on device >=20 > Rob >=20 >=20 > On Jan 25, 2012, at 9:41 AM, Donovan Brooke wrote: >=20 >> First, I wouldn't call something a bug unless I can confirm and = reproduce it. Why exactly are "we" calling this a bug in 6.2? How is it = reproduced? >>=20 >> To answer below: >>=20 >> [snip] >>=20 >> Also, we don't know if there are other revisions or fixes Donovan = team may have done for 6.2 that aren't in our SVN. >>>>=20 >>>> I suggest you speak privately (or publicly) with Rob Forest who has = the same problem with his 6.2 copy: a group of several users could club = together to get it fixed. Our charge would be somewhere between $800 to = $1200 >>=20 >>=20 >> I think Chris knows that I'd be interested in a build/patch to = CICADA.. not for the stated "bug" really, but for the Apache2 fixes that = we made to the plugin (content type text/XML), and the SMTP auth fixes. = Those should both already be done in the 6x branch and awaiting a build. >>=20 >> There may be some other fixes that were made, and there may be some = loose end's as well... since there was some "changes" going on with WSC = policy about the time 6x was tagged and when I left. I don't know = exactly what state 6.2 was tagged at, but it should be doable without = too much problem. The developer would want to make sure they use the = apache module with the fix on the content type. >>=20 >> We were also playing around with process isolation.. much like how = WebDNA7 works. WebDNA 6.2 does not have to be installed server-wide and = we did have a build working that would allow multiple installs on a = single server. Not sure if this was kept or not. >>=20 >> Anyway... ahh, memories. :-) >>=20 >> Donovan >>=20 >>=20 >> --=20 >> Donovan Brooke >> Euca Design Center >> www.euca.us >>=20 >> --------------------------------------------------------- >> 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 >=20 > --------------------------------------------------------- > 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
christophe.billiottet@webdna.us
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:
I can't set the variable (2005)
WebCat2 - [include] tags (1997)
[include ...] behavior (1997)
Re:2nd WebCatalog2 Feature Request (1996)
Export database record - append to other database (2005)
Re:no template caching (1997)
[WebDNA] checking for mixed case text (2009)
How can I record purchases to a database? (1998)
WebCatalog2 Feature Feedback (1996)
[WebDNA] High-profile WebDNA sites? (2008)
[date] formatting bug inside [orderfile]? (1998)
[WAY OFF TOPIC] anybody live in Kansas? (2003)
Sendmail strangie (2002)
&fieldsdir=ra truely random?? (2000)
Categories and Sub-categories (2003)
how to put/keep product customization info in basket? (1998)
no global [username] or [password] displayed ... (1997)
New 4.5 installer (2002)
French Webcat User (2002)
pass a captured value (2004)