PHP on UNIX15

[dpyers]

For windows boxes, enter a fake subdomain into the browser and it returns the server name. Nothing comparable for unix that I know of.

[jdupton]

If one doesn't currently have any sites hosted on unix15, (I don't) is there a straightforward way to create a temporary account on a specific server to test sample site PHP code?

[admin]

The only thing that changes is php no longer runs or writes as httpd it now writes as the account user name. This way we can also track those bad scripts like fopen.

The testing has been going on for a about a week on a non-live server, there was no delay in letting you guys know, it was not fully live on the system until after we posted and then we sent the email after the post, bit after the post. But I got a bit busy today that was my fault. But we have seen zero issues in this change so far other then the odd mail() thing.

BTW paulyc I rasied hell at the tech in charge of this because I thought the same thing that it had something to with the other. He found the real issue and need to say I had to tell him sorry for bitching him out for letting something like slip.

Things this will do for us and you.

1. Better security because we will know what whos php's files are being called.
2. We will know when hackers try and call perl from php to run DDOS attacks.
3. Your files will be writen as the user now httpd.
4. PHP now uses less memory as well. Down to 100MB from 150MB. ( this was one thing we could not fully test but was sure it would not case any issues since we had more ram then needed )
5. CPU load is about the same 0.5 to 0.8 or less. Those numbers are from peak ET.

So lots of good things and have seen

[dvanburen]

This point:

3. Your files will be writen as the user now instead of httpd.

is very good news for anyone new trying to setup a CMS. I can't count how many times I had to reset permissions on files/directories because they were owned by httpd.

[paulyc]

Originally Posted by tracewebspace

thought so.

what was my point you ask?

these are sharted servers, anyones script can cause the server to come to a halt, and while they are generally reliable, if you have a site that needs to be online 24 x 7 and only down when you take it down for maintenance, etc, get yourself a dedicated server. They are as low as $99 a month. Would even suggest vortech's vps's. lot more reliable than a shared server and nobody elses scripts or problems will bring your site down.

Does price make a differentce in the quality of hosting, yes! While I dont believe there is a better value out there than vortechs offerings, dont kid yourself that they will be up 24 x 7, that there will not be issues, that they wont upgrade php, etc which will cause an issue somewhere in someones script, etc, etc. Thats the trade off you get when you save a few $$.

Gee Wiz - I stand corrected chicago...

... <dont_remember> asking you what your point was </ dont_remember>

...

[paulyc]

Originally Posted by ferretman

Didn't your issue occur before Unix15 was modified?

You said the issue took 2 days to resolve and the first mention of work on Unix15 was at 8:15 this morning.

Brad, did the support team delay telling us or were the changes made this morning?

<chuckle>
So... what.... you guys really think like... magically at 11:15 AM (unless I missed some pre post) - *poof* - new php.
</chuckle>

surely you realize you got the public announcement at 11:15 but work had been going on for ... hours? - days?

I recieved official notice via email from my client at 8:30 am Monday Morning -
I spent 2 hours exhausting all possibilities and making sure it was in fact an issue with a php function call before opening a ticket at 11:30 am


This is an excerpt verbatim from a Tech Support response to me at around 1:00 pm on Monday June 20
************************************************** ********
I know that one of our technicians is testing a new PHP installation (on a seperate webserver and port) on Unix15, but he has advised that it should not effect Mail() function of the existing build of PHP on the standard server (port 80).
************************************************** ********

My cients issue was resolved this morning sometime before noon -

the reason? (I always ask)
-------
Sendmail was failing due to the lack of inodes available to the system.
-------

is it possible that inodes were tied up helping out a busy php install? I don't know enough about it - matter of fact I know nothing about inodes or unix. don't have to thank god. I'll stick to actionscript and other scripting languages thank you very much.

So - ok 2 days was an exaggeration - but I was a bit perturbed -- when you have an angry customer losing money - an hour seems like a day.

Just for the record - over the last 5 years I have been through a few hosting providers, from expensive ones to cheap ones. I have always been relatively satisfied with the quality of the hosting at vortechmatrixrapidcolonewconneticutcompanyiforgett henameofvixen hosting. Considering the cost - you almost can't beat it. I have always found the support very responsive - albeit not always 100% knowledgable and on point - but very good response times none the less and one way or the other the issue gets resolved.

I just get cranky when the car dies for no damn reason and the baby starts screaming cause we're stuck on teh side of the road not moving - ya know?

[orbitrus]

Originally Posted by paulyc

is it possible that inodes were tied up helping out a busy php install?

Not really. inodes are used in a table based filesystem (XFS, EXT, JFS, etc) - hence tree filesystems (Reiser, etc) are unaffected. Each file uses an inode.

In the case of /var being filled, that's because (mainly) /var holds:
/var/log
/var/tmp
/var/spool/mail
/var/www

There is a tradeoff between inodes and disk space, as the more inodes you have, the less space (the table takes space).

Anyways, running out of inodes is like running out of disk space. There is no room left for the file.

...and a busy PHP install would use /tmp and /usr, not /var in general, busy creating about 400 files [inode # are usually quite high, eg my 2GB / has 275904 total, 160642 free]

Originally Posted by paulyc

I don't know enough about it - matter of fact I know nothing about inodes or unix. don't have to thank god. I'll stick to actionscript and other scripting languages thank you very much.

Shame. Unix is quite interesting (and are languages like, oh, ASM, C, Java, .NET). But then, I admit I like PHP/Javascript still.



BTW Brad: On the advise of someone on your Tech support staff [who said it wouldn't be that tough] I tried out Gentoo [up from RH9] about 8 months ago. Still running it. Thanks for the encouragement to get out and try something new! (And while I was at it learn C/C++ as my University classes only tought Java)

[Trusun]

Thx admin your tech must have fixed the problem cuz everything is working now.. thx

[tkraffty]

I'm hoping these changes weren't also applied to other servers. My client on NT18, windows PHP is using an industry-standard content management application (MT - or MoveableType) which uses perl, mySQL and perl/cgi to produce site files. He reported the exact same "sendmail not found" error to me the other day, and just today I found out they're also getting message to the effect of "posted item saved, but errors exist that prevent pinging" (MT provides content syndication, which along with email pings and notifications was working fine for two months until just last week)

[paulyc]

To be honest this is my biggest pet peeve with Vortech - too often stuff like this will magically stop working and the first you hear of it is from a frustrated client.

[admin]

Originally Posted by tkraffty

I'm hoping these changes weren't also applied to other servers. My client on NT18, windows PHP is using an industry-standard content management application (MT - or MoveableType) which uses perl, mySQL and perl/cgi to produce site files. He reported the exact same "sendmail not found" error to me the other day, and just today I found out they're also getting message to the effect of "posted item saved, but errors exist that prevent pinging" (MT provides content syndication, which along with email pings and notifications was working fine for two months until just last week)

Hmm this is unix only not windows. Windows already does what suphp does for unix.

[admin]

Originally Posted by paulyc

To be honest this is my biggest pet peeve with Vortech - too often stuff like this will magically stop working and the first you hear of it is from a frustrated client.

If you upgrade your forum account you might see we are trying to inpove this and the time to notify customers. You are still only a registered user.

I just told all techs before changes go live on a system customers must be notified with as much notice as we can give. In some case we can't give is much as you or I would like. But security comes very high on our list.