My host is upgrading to PHP/SQL5. Do I need to do anything/worry?

[青山 素子]

Mouthrax, is it causing 500 errors, or just not working? Those can be two different problems.

[weightman]

I am with HostGator. They haven't switched me over yet though.

[Mouthrax]

Not with Hostgator, I am with streamline.net

It is giving the 500 internal error message, also no post reply notifications are being sent out?

I am just at a loss, I am sure it is the server side but you know what they are like, they will try blame apps rather than admit it is a problem there side so just want to have some insite to reply to them if I get fobbed off.

So far they have always been great and sorted any probs out.

[青山 素子]

Unfortunately, the server error logs are really the only way to determine the cause of the error. If your host uses cPanel, you should be able to view the logs. If they are using something else it is possible you also have access to view those, just look around.

[Mouthrax]

I have found an error log that has lots of errors from yesterday and today which is the times in question.

The file is full of stuff like this from different IP's

[Thu Mar  6 23:16:04 2008] [error] [client XX.XX.XXX.XXX] FastCGI: comm with (dynamic) server "/var/www/fcgi/php-cgi" aborted: (first read) idle timeout (30 sec)
[Thu Mar  6 23:16:04 2008] [error] [client XX.XX.XXX.XXX] FastCGI: incomplete headers (0 bytes) received from server "/var/www/fcgi/php-cgi"

Any help?

[青山 素子]

That seems to indicate some problems with PHP itself or possibly FastCGI.

[Mouthrax]

That seems to indicate some problems with PHP itself or possibly FastCGI.

Sorry does that mean server side? ...and will that have to do with the update to php5?

[青山 素子]

It means PHP itself is possibly having problems. You usually see this type of error when there is a problem with one of the server components.

Basically, your server appears to be set up as "Apache <-> FastCGI <-> PHP". Something is going wrong in the connection between those items, most likely between FastCGI and PHP. It could be anything from the wrong PHP command being used (the CLI instead of the CGI, or the CGI didn't have FastCGI support enabled when built) to a configuration issue with FastCGI.

[TheEmperor]

Not with Hostgator, I am with streamline.net [nofollow]

It is giving the 500 internal error message, also no post reply notifications are being sent out?

OK this is exactly the same problem I have: We are on Streamline.net [nofollow], we are getting a lot of 500 errors and notifications stopped (along with password retrieval emails).

As we've done the whole Streamline route and got replies from the moderately helpful (but not to the point of fixing things) to the random (copy and paste) to the unhelpful I'll save you that other.

What I'll do is break the problem down:

• This started when we got them to move us to a PHP 5 server from a PHP 4.
• When that happened all the forum notifications went and so did blog notifications and the contact form stopped working.

This would suggest to me problems with their mail server settings, except....

• This just happened to another SMF forum I visit and they aren't on Streamline, yet their contact forms have also gone wrong.
  
• When I tried to login to reply to this I couldn't remember my password and requested a new one - which has never shown up (I've had to start a new account). Are there problems here?
• The SMF user registration password works just fine here and on our forum - the email turned up in the inbox within minutes.

So it almost suggests problems with the mail server and also almost suggests issues with SMF but not quite (the exceptions seeming to discount either explanation).

I found other more recent threads about the notifications suddenly stopping working:

www.simplemachines.org/community/index.php?topic=229630.0
www.simplemachines.org/community/index.php?topic=228873.0

This last one seems to suggest you need to update the SMF mail script to work with PHP 5. Is this right? Why would the new registration email work but the notifications and password reminder fail?

I'll start noddling on this but if anyone can throw in clear instructions on what needs to be changed and where that'd certainly help me and the other people struggling with this issue.

Thanks in advance for any help you can provide.

[青山 素子]

This last one seems to suggest you need to update the SMF mail script to work with PHP 5. Is this right? Why would the new registration email work but the notifications and password reminder fail?

AFAIK, the mail() function in PHP hasn't changed since 4.3.0 and 4.2.3 was the last significant change (with additional parameters being blocked in safe_mode).

Edit: I just checked the reminder function with SMF 1.1.4 on my dev server (Apache 2.2.6 with PHP 5.2.4 on Linux - yes, I know I need to update) and it sent just fine. I am using the PHP default method.

Mar 21 11:43:12 **** postfix/pickup[9165]: DC63FDDEC9: uid=80 from=<apache>
Mar 21 11:43:12 **** postfix/cleanup[9390]: DC63FDDEC9: message-id=<20080321184312.DC63FDDEC9@********************>
Mar 21 11:43:12 **** postfix/qmgr[3178]: DC63FDDEC9: from=<apache@********************>, size=1330, nrcpt=1 (queue active)
Mar 21 11:43:13 **** postfix/smtp[9392]: DC63FDDEC9: to=<***********@*****************>, relay=***.***.***.***[***.***.***.***]:25, delay=0.34, delays=0.18/0.02/0.07/0.07, dsn=2.0.0, status=sent (250 2.0.0 Ok: queued as 26402934F8)
Mar 21 11:43:13 **** postfix/qmgr[3178]: DC63FDDEC9: removed

[TheEmperor]

Thanks for the help - I took it over to the other thread and did some fiddling and seem to have fixed the problem, which should also fix Mouthrax's problem (and possibly others with the same issues):

www.simplemachines.org/community/index.php?topic=228873.msg1477641#msg1477641

Thanks again for the help.

[Mouthrax]

Testing fix now

[Mouthrax]

No still getting 500 Internal Errors

[Mouthrax]

...Although the forum does seem more stable than it was, I have only had one error message in 10 mins.

Could the fix need applying elsewhere?

[SleePy]

Have you asked your host?

That error is not generated by smf but the system itself.

[weightman]

So HostGator finally made the upgrade, without sending me the advanced email they promised. I can access the forum and it seems to be working properly at its Mambo/Orstio bridge address: http://weightliftingexchange.com/index.php?option=com_smf&Itemid=39
but my home page, which uses ssi, is giving the error listed below now. Can anyone help? Should I ask Orstio in the mambo bridge board?  Thanks

Fatal error: Cannot redeclare ssi_shutdown() (previously declared in /home/xxxxxx/public_html/smf/SSI.php:191) in /home/xxxxxx/public_html/smf/SSI.php on line 193

[SleePy]

weightman,

Did you solve this?

That means SSI was included twice.
You would need to check the source code to see if you have maybe accidently included SSI.php more than once.

[weightman]

Yes, pretty much solved. I had multiple SSI modules on my SMF bridged home page. I never did figure out how to get the two modules to work together so I removed one- something to do, as you said, with require, require_once_, include, include_once or something probably- no biggie.

[Sarge]

Yes, pretty much solved.

Glad the upgrade was smooth for you! I'm marking this topic as solved.

I had multiple SSI modules on my SMF bridged home page. I never did figure out how to get the two modules to work together so I removed one- something to do, as you said, with require, require_once_, include, include_once or something probably- no biggie.

That's what require_once() is supposed to do:

require_once() should be used in cases where the same file might be included and evaluated more than once during a particular execution of a script, and you want to be sure that it is included exactly once to avoid problems with function redefinitions, variable value reassignments, etc.

I'm not an SMF dev, but it seems that require_once() is the preferred way to include files in SMF.

[weightman]

Yea, I read that. But, it didn't work. It could be that require is in yet another SSI module but if so, then more than one has the require statement right now and is working. I sort of gave up pretty quick though because it wasn't all the big a deal.