Re: [WebDNA] v7 problem?
This WebDNA talk-list message is from 2012
It keeps the original formatting.
numero = 108008
interpreted = N
texte = > 70,000 hits a day is by no mean a "high load". >=20 > Several questions to help us debug this: >=20 > - you write "So I have 3 v7 hosts running on one server and another 2 = on the second server. As soon as I add another one the server crashes" = which could imply a limit in the number of running FastCGI processes. = However, you also wrote ""We've got 4 vhosts running v7 since a few = weeks without problems". It's definitely 3 vhosts on server 1 (MacPro dual six-core) and 2 vhosts = on server 2 (MacPro dual quad-core) and they are running fine like this. > - Is it always the same website that makes your webserver crash? The "error" appears as soon as I add another vhost with DNA7 on any of = the two servers. The vhost I add contains ONLY a WebDNA Folder and is NOT hit by clients. > - when you write "the server crashes", do you mean the website = (WebDNA), apache (all websites) or the entire server? I get two kinds of errors in the apache error log as soon as I do = apachectl graceful: - sem_open failed: returned EEXIST: File exists this on is repeated about 20 times and the server still works - retry on sem_wait: Bad file descriptor This error gets repeated endless and unbelievable fast, so that the = logfile grows 1 GB per Minute till the disk is full! (it ended up with a = size of 480GB) The server did not really crash. > - Do you know if there is a specific apache/iTools directive that = could limit the child processes and automatically restart them when it = reaches its limit? it could be that the iTool apache config is trying to = spawn a high number of childs and that there is a limit somewhere else, = at the mod_FastCGI config level >=20 > to simplify debugging, you could try > StartServers 1 > MinSpareServers 1 > MaxSpareServers 1 Currently we have the following values: Start Servers: 5 Max Clients: 150 MaxSpare Servers: 10 MinSpare Servers: 5 Max Requests per Child: 0 (unlimited) Basically iTools just manages apache config files, but I didn't find = anything mod_FastCGI specific. Before changing this values I'll contact Tenon and ask them ... Thanks Thierry >=20 > - chris >=20 >=20 > On Jan 9, 2012, at 10:33, Thierry Almy wrote: >=20 >> All of our sites are not "extremely" busy ... a few ones have more = hits than other that's true. >>=20 >> For server 1 we had about 70'000 hits per day (hits, not visits!) = during the last two weeks, but all served with v6.2 >> The ones using v7 are not heavily hit. >>=20 >> Server 2 has maybe half of hits on v6.2 and only two v7 sites with = almost no hits. >> The "www.maendli-freizeit.ch" vhost is in development state and has = no hits at all, as I mentioned it makes them crash also if it contains = only the WebDNA folder and nothing else. >> It doesn't really matter if the v7 sites are heavily hit or not, it's = the fact that having more than a handfull of fcgi processes running at = once. >>=20 >> Thierry >>=20 >>=20 >> Am 09.01.2012 um 13:20 schrieb christophe.billiottet@webdna.us: >>=20 >>> Yes, we will check this error and i will be back with informations. = Could it be that you are running extremely busy websites on a single = server, the kind of website that usually run on dedicated servers, and = that one is heavier than the others? i understand the error follows one = of your websites, www.maendli-freizeit.ch, and that the problem appears = when you add this specific website to your servers. >>>=20 >>> I also understood the problem appeared Saturday: "We've got 4 vhosts = running v7 since a few weeks without problems - till yesterday." >>>=20 >>> - chris >>>=20 >>>=20 >>> On Jan 9, 2012, at 10:07, Thierry Almy wrote: >>>=20 >>>> Thanks Chris >>>>=20 >>>> On our 2x QuadCore we can handle 2 fcgi processes, on the 2x 6-Core = 4 fcgi processes.=20 >>>> The next process will bring up the errors on each system. >>>>=20 >>>> To be honest, it can't be a solution to set up a VPS for each web = host. We're running more than 100 vhosts, it's just impossible to do it = this way ... >>>>=20 >>>> The sem_open error points to a programming problem where the = application doesn't properly close channels after usage ...as far as I = understand it ... but I'm definitely not an application programmer ....=20= >>>>=20 >>>> Thierry >>>>=20 >>>>=20 >>>> Am 09.01.2012 um 12:19 schrieb christophe.billiottet@webdna.us: >>>>=20 >>>>> I will check this asap with our developer. Meanwhile, not all = hosting solutions are being equals, though the FastCGI should be able to = handle almost any UNIX/LINUX configuration. I personnally prefer the one = chosen by Ken: a single website on a VPS with a light but feature-rich = linux OS like DEBIAN associated with LIGHTTPD webserver -which is faster = and lighter than apache- and has a well-designed FastCGI implementation. = This solution can be very cheap, will run with 128MB of RAM and scales = up easily with fast processors and lots of RAM. >>>>> The VPS is the natural extension of WebDNA 7: it gives as much = control as a dedicated server for a fraction of the cost, and makes each = website an independant entity with greater efficiency and high level of = security. >>>>>=20 >>>>> - chris >>>>>=20 >>>>>=20 >>>>> On Jan 9, 2012, at 8:39, Thierry Almy wrote: >>>>>=20 >>>>>> Hi chris >>>>>>=20 >>>>>> sorry for the comment ... I take it back ... >>>>>>=20 >>>>>> I just want to have an environment which allows me to do my job = (programming). >>>>>>=20 >>>>>> Maybe we should leave OS X at all and move to Linux? >>>>>>=20 >>>>>>=20 >>>>>> The last error we had a few weeks ago was because of heavy load = and bad RAM. >>>>>> After changing the RAM the malloc error was gone. >>>>>> The sem_open error must have to do something with WebDNA ... v6.2 = and v7 >>>>>>=20 >>>>>> Is there anybody else which has seen the sem_open error? >>>>>>=20 >>>>>> Thierry >>>>>=20 >>>>>=20 >>>>>=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 >>>=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 >=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:
> 70,000 hits a day is by no mean a "high load". >=20 > Several questions to help us debug this: >=20 > - you write "So I have 3 v7 hosts running on one server and another 2 = on the second server. As soon as I add another one the server crashes" = which could imply a limit in the number of running FastCGI processes. = However, you also wrote ""We've got 4 vhosts running v7 since a few = weeks without problems". It's definitely 3 vhosts on server 1 (MacPro dual six-core) and 2 vhosts = on server 2 (MacPro dual quad-core) and they are running fine like this. > - Is it always the same website that makes your webserver crash? The "error" appears as soon as I add another vhost with DNA7 on any of = the two servers. The vhost I add contains ONLY a WebDNA Folder and is NOT hit by clients. > - when you write "the server crashes", do you mean the website = (WebDNA), apache (all websites) or the entire server? I get two kinds of errors in the apache error log as soon as I do = apachectl graceful: - sem_open failed: returned EEXIST: File exists this on is repeated about 20 times and the server still works - retry on sem_wait: Bad file descriptor This error gets repeated endless and unbelievable fast, so that the = logfile grows 1 GB per Minute till the disk is full! (it ended up with a = size of 480GB) The server did not really crash. > - Do you know if there is a specific apache/iTools directive that = could limit the child processes and automatically restart them when it = reaches its limit? it could be that the iTool apache config is trying to = spawn a high number of childs and that there is a limit somewhere else, = at the mod_FastCGI config level >=20 > to simplify debugging, you could try > StartServers 1 > MinSpareServers 1 > MaxSpareServers 1 Currently we have the following values: Start Servers: 5 Max Clients: 150 MaxSpare Servers: 10 MinSpare Servers: 5 Max Requests per Child: 0 (unlimited) Basically iTools just manages apache config files, but I didn't find = anything mod_FastCGI specific. Before changing this values I'll contact Tenon and ask them ... Thanks Thierry >=20 > - chris >=20 >=20 > On Jan 9, 2012, at 10:33, Thierry Almy wrote: >=20 >> All of our sites are not "extremely" busy ... a few ones have more = hits than other that's true. >>=20 >> For server 1 we had about 70'000 hits per day (hits, not visits!) = during the last two weeks, but all served with v6.2 >> The ones using v7 are not heavily hit. >>=20 >> Server 2 has maybe half of hits on v6.2 and only two v7 sites with = almost no hits. >> The "www.maendli-freizeit.ch" vhost is in development state and has = no hits at all, as I mentioned it makes them crash also if it contains = only the WebDNA folder and nothing else. >> It doesn't really matter if the v7 sites are heavily hit or not, it's = the fact that having more than a handfull of fcgi processes running at = once. >>=20 >> Thierry >>=20 >>=20 >> Am 09.01.2012 um 13:20 schrieb christophe.billiottet@webdna.us: >>=20 >>> Yes, we will check this error and i will be back with informations. = Could it be that you are running extremely busy websites on a single = server, the kind of website that usually run on dedicated servers, and = that one is heavier than the others? i understand the error follows one = of your websites, www.maendli-freizeit.ch, and that the problem appears = when you add this specific website to your servers. >>>=20 >>> I also understood the problem appeared Saturday: "We've got 4 vhosts = running v7 since a few weeks without problems - till yesterday." >>>=20 >>> - chris >>>=20 >>>=20 >>> On Jan 9, 2012, at 10:07, Thierry Almy wrote: >>>=20 >>>> Thanks Chris >>>>=20 >>>> On our 2x QuadCore we can handle 2 fcgi processes, on the 2x 6-Core = 4 fcgi processes.=20 >>>> The next process will bring up the errors on each system. >>>>=20 >>>> To be honest, it can't be a solution to set up a VPS for each web = host. We're running more than 100 vhosts, it's just impossible to do it = this way ... >>>>=20 >>>> The sem_open error points to a programming problem where the = application doesn't properly close channels after usage ...as far as I = understand it ... but I'm definitely not an application programmer ....=20= >>>>=20 >>>> Thierry >>>>=20 >>>>=20 >>>> Am 09.01.2012 um 12:19 schrieb christophe.billiottet@webdna.us: >>>>=20 >>>>> I will check this asap with our developer. Meanwhile, not all = hosting solutions are being equals, though the FastCGI should be able to = handle almost any UNIX/LINUX configuration. I personnally prefer the one = chosen by Ken: a single website on a VPS with a light but feature-rich = linux OS like DEBIAN associated with LIGHTTPD webserver -which is faster = and lighter than apache- and has a well-designed FastCGI implementation. = This solution can be very cheap, will run with 128MB of RAM and scales = up easily with fast processors and lots of RAM. >>>>> The VPS is the natural extension of WebDNA 7: it gives as much = control as a dedicated server for a fraction of the cost, and makes each = website an independant entity with greater efficiency and high level of = security. >>>>>=20 >>>>> - chris >>>>>=20 >>>>>=20 >>>>> On Jan 9, 2012, at 8:39, Thierry Almy wrote: >>>>>=20 >>>>>> Hi chris >>>>>>=20 >>>>>> sorry for the comment ... I take it back ... >>>>>>=20 >>>>>> I just want to have an environment which allows me to do my job = (programming). >>>>>>=20 >>>>>> Maybe we should leave OS X at all and move to Linux? >>>>>>=20 >>>>>>=20 >>>>>> The last error we had a few weeks ago was because of heavy load = and bad RAM. >>>>>> After changing the RAM the malloc error was gone. >>>>>> The sem_open error must have to do something with WebDNA ... v6.2 = and v7 >>>>>>=20 >>>>>> Is there anybody else which has seen the sem_open error? >>>>>>=20 >>>>>> Thierry >>>>>=20 >>>>>=20 >>>>>=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 >>>=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 >=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
Thierry Almy
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:
[BULK] [WebDNA] [BULK] Mac OS X LION has no FastCGI (2011)
Credit Card Number Stripping Digit (1997)
Stopping bad HTML propagation ? (1997)
WebCatalog Mac and cgi-bin (WebSTAR 2.0) (1997)
ANother SHOWIF problem (1997)
Fwd: Problems with Webcatalog Plug-in (1997)
free "Image Viewer" widget (2004)
Secure Web Server (1999)
back button loses cart (2000)
Boolean showifs? (2000)
too many nested tags ... (1997)
Re:Help name our technology! (1997)
possible, WebCat2.0 and checkboxes-restated (1997)
Site Search Concepts (2003)
WebCat2b12--[searchstring] bug (1997)
Sorting LISTFILES (1997)
WCS Newbie question (1997)
Multiple Merchant Accounts? (1997)
Shownext! (1998)
[WebDNA] Webcat 6 - MacIntel - iTools (2008)