Linux Shared Hosting woes with Web Africa (again)

[TheRealWitblitz]

Okay so I'm a home user and we use WA's hosting services for a couple sites I did for family members to showcase their small businesses.

For the one site in particular we've used their SA Premium Shared Linux Hosting service for the past year, which has been running 100% (for the most part). It's based on Wordpress with a custom theme and just a couple plugins, so no real overhead as I mentioned the site has been fine.

But about a week ago (without my knowledge maybe they are too busy or I'm too small to bother with...), the site returned a http 500 error. I opened a ticket and they notified me that my site has, in their words :

"your domain has "out grown" our shared environment and would now be requiring more resources than is allowed to one particular domain at any given time.

The domain would need to utilize one of our more dedicated solutions such as a VPS solution which is dedicated resources which you setup as required"

So I figured how is this possible I haven't inflated the database beyond its allowed size (it's only 5mb), I haven't done any code changes, I didn't even add new plugins for over a year. I checked the CPanel stats the site is using virtual memory of around 184mb/1024mb, disk space well within limits of allotted 5Gb.

So I made backups of the files/dbase. And proceeded to delete all the files and wiped the database clean. I re-uploaded a simple index.php to the root of the site and low and behold I still get the error. With all the supposed strain removed from their server how am I now using up all the allowed resources if they can't even deliver a simple php page, yet their support desk seems adamant that I need a VPS?

The only indication is from the log :

[Wed Sep 25 17:47:59 2013] [error] [client 41.xx.xx.xx] (12)Cannot allocate memory: couldn't create child process: /opt/suphp/sbin/suphp for /home/<user_withheld>/public_html/index.php

What gives? I cleared all the logs as well. Seems to me like a misconfiguration or permission issue regarding my user ID on their part.

 

[TheRealWitblitz]

The thing that infuriates me the most is the fact they can't even fully look into the problem, it's their server after all they are in the business of hosting so they have all the expertise and know-how, and have the qualified people who can give me, the small home user the solution. Yet now it falls to me to ****ing research and provide feedback in order for them to fix their own problems.

From my p.o.v. I should be able to FTP a basic working php file and at least get that served back to me without erroneous errors, yet this back and forth with them has been going on for nearly a week. Inexcusable to be honest.

 

[w1z4rd]

Looks like your site is chowing RAM and theyre not managing it all that well. Technically the amount of ram you use should be managed so your website does not impact other shared hosting clients. Since theyre not doing that your site is probably impacting their other clients.

So if you want to stayed in the shared hosting environment I would recommend a company that manages their shared hosting differently (I personally recommend domaincheap.co.za) or I would upgrade to a VPS (if your site is doing a lot of traffic).

There are howerver, limits to shared hosting. While your site might be small a hacker could have uploaded a script that uses a ton of resources. Size of site does not translate very well to how more resources it uses. Sometimes a small site can use a ton of resources and sometimes a big site can use very little. It depends on how its written and what it does.

 

[TheRealWitblitz]

I'm spiking 1Gb of ram by running a basic php file with contents : <?php echo 'This is a simple php test.'; ?>

You're telling me that single line is hammering their server so hard I'll need that VPS after all. Hah!

 

[w1z4rd]

I'm spiking 1Gb of ram by running a basic php file with contents : <?php echo 'This is a simple php test.'; ?>

You're telling me that single line is hammering their server so hard I'll need that VPS after all. Hah!

How dont know how they manage their resource usage. Perhaps its accumulates over time or something. You shouldn't be spiking a GB of ram for that.

Do you spike when you empty your hosting folder? Like when you are sure there are no infected plugins pwning cpu usage?

 

[TheRealWitblitz]

Like I said I deleted everything in the public_html folder, I left the disk usage bare only an index.php file sits in the root. I wiped the database clean as well. Cleared the logs. The email accounts on the domain doesn't even have 100mb of data between them.

 

[TheRealWitblitz]

Here are some server status stats :


General Server Stats
rsyslogd up
pop up
clamd up
ftpd up
sshd up
lfd up
imap up
exim (exim-4.80.1-1.cp1136.x86_64) up
cpdavd up
mysql (5.5.32-cll) up
httpd (2.2.25 (Unix)) up
dnsadmin up
cpsrvd up
ipaliases up
crond up
Server Load 4.24 (4 cpus)
Memory Used 57.8 %
Swap Used 1.05 %
Disk /dev/sda3 (/) 40 %
Disk /usr/tmpDSK (/tmp) 3 %
Disk /dev/sda2 (/usr) 41 %
Disk /dev/sda1 (/boot) 74 %
Disk /dev/sdb1 (/home) 51 %
Disk /tmp (/var/tmp) 3 %
Disk /dev/sdc1 (/var/lib/mysql) 21 %

Personal Stats
CPU Usage 0 / 100 %
Virtual Memory Usage 189,536 / 1,048,576 KB
Entry Processes 0 / 15
Disk Space Usage 97.09 / 5,120 MB
MySQL Databases 1
MySQL Disk Space 0 MB
Hosting package SA Premium
Server Name lnxwebs01

 

[TheRealWitblitz]

And some resource usage graphs for the past 7 days :

 

[Solitude]

That really doesn't look to me like your site has outgrown a shared environment. I would love to know what resources you have been using too much of. It's rather disappointing that their reply to your query has been answered so lazily and without much detail.

I look forward to a Web Africa representative giving some more information in this thread.

 

[TheRealWitblitz]

Its such a basic site even with low traffic. Surely they don't set the standards THAT low? It's not the bottom of the range hosting option either. Very confusing.

 

[guest2013-1]

Someone else on the shared server is ****ing up. The 500 error you're getting looks like the server itself couldn't spool another httpd process for you (or hit an internal limit of processes). I know hostgator terminates any processes waiting in queue if there is more than 25, so someone else must be hogging it, and knowing Web Africa (who can't host a nude convention on a nude beach to save their lives), they probably changed a setting on the server to "make it better" and screwed the entire server.

Good luck licking their ass to find out and fix the problem permanently. You'd be better off moving to someone like elitehost

Also, Wordpress isn't the most friendly to resources, but with what you're using it for, this shouldn't be a problem at all unless Google hits you like a pimp for some reason

 

[Webafrica Helper]

Okay so I'm a home user and we use WA's hosting services for a couple sites I did for family members to showcase their small businesses.

For the one site in particular we've used their SA Premium Shared Linux Hosting service for the past year, which has been running 100% (for the most part). It's based on Wordpress with a custom theme and just a couple plugins, so no real overhead as I mentioned the site has been fine.

But about a week ago (without my knowledge maybe they are too busy or I'm too small to bother with...), the site returned a http 500 error. I opened a ticket and they notified me that my site has, in their words :

"your domain has "out grown" our shared environment and would now be requiring more resources than is allowed to one particular domain at any given time.

The domain would need to utilize one of our more dedicated solutions such as a VPS solution which is dedicated resources which you setup as required"

So I figured how is this possible I haven't inflated the database beyond its allowed size (it's only 5mb), I haven't done any code changes, I didn't even add new plugins for over a year. I checked the CPanel stats the site is using virtual memory of around 184mb/1024mb, disk space well within limits of allotted 5Gb.

So I made backups of the files/dbase. And proceeded to delete all the files and wiped the database clean. I re-uploaded a simple index.php to the root of the site and low and behold I still get the error. With all the supposed strain removed from their server how am I now using up all the allowed resources if they can't even deliver a simple php page, yet their support desk seems adamant that I need a VPS?

The only indication is from the log :

[Wed Sep 25 17:47:59 2013] [error] [client 41.xx.xx.xx] (12)Cannot allocate memory: couldn't create child process: /opt/suphp/sbin/suphp for /home/<user_withheld>/public_html/index.php

What gives? I cleared all the logs as well. Seems to me like a misconfiguration or permission issue regarding my user ID on their part.

Hi,

I can't provide any info around the issue as yet as I will need to get your customer code and the website name so I can take a look. If you provided all this info to our support team they should have investigated it properly.

Please PM me your details so I can take this further.

 

[guest2013-1]

Hi,

I can't provide any info around the issue as yet as I will need to get your customer code and the website name so I can take a look. If you provided all this info to our support team they should have investigated it properly.

Please PM me your details so I can take this further.

stop talking bull****. To open a support ticket you need to be registered. He clearly said:

I opened a ticket

Clearly this issue has been logged and the sales team you hire as engineers saw an opporunity to pawn off a higher hosting package to him instead of actually looking at the issue

 

[Webafrica Helper]

stop talking bull****. To open a support ticket you need to be registered. He clearly said:



Clearly this issue has been logged and the sales team you hire as engineers saw an opporunity to pawn off a higher hosting package to him instead of actually looking at the issue

Hi AcidRazor,

If you read 'TheRealWitblitz' post he never once puts any of his personal details in the post, nor does he put his customer code or the ticket reference number in the post. So how do I know what his account details are? Clearly not from his post in this thread, so I have to ask for it.

I would have not asked him for his details if I had them already and I never said he is not a customer. I said I cannot provide any info on his complaint as I need to get his details to take a look and then I can take it from there. Another point to mention is that I never said that our Hosting team handled the matter correctly and I believe it could have been handled much better than it has been.

I think your outright hatred for Web Africa is clouding your judgement on matters like these but nonetheless 'TheRealWitblitz' has pm'd me his details a while back and we are in the process of getting things sorted.

 

[TheRealWitblitz]

Issue resolved for the time being. Ty for the assistance. They'll be monitoring usage.

 

[Solitude]

Issue resolved for the time being. Ty for the assistance. They'll be monitoring usage.

What happened in the end? Eh, you can't just give us all that information in the beginning and then end the thread like that.

 

[RSkeens]

This looks to be quite a standard CloudLinux error, if that is what they use then it more than likely involved them tweaking the user's resource allowance.

 

[TheRealWitblitz]

Well... this morning everything was fine it finally loaded phpinfo();. I optimised the database and updated to the latest WP 3.6.1 build, and updated all the plugins on my local dev machine before re-uploading the files and database to the server.

The reason I was given :

Under normal circumstances and for physical memory our limits are generous and protects customers from feeling the effect of 1 site over-utilizing the resources, but in this case our limit for virtual memory was too strict and we have now increased the limit substantially.

But personally why would php pack up and go all together, unless they have something in place to keep my site in lock-down once I hit the thresholds. But I'm no expert on server architecture, linux, etc.

They know the reasons, and as long as my site is working within acceptable limits I'm happy. I didn't need that VPS after all, or so it seems. It's a small site.

 

[Solitude]

Thanks for the update. I'm glad that it was sorted out, though it is disappointing that you had to post here before they helped you properly.

Sometimes I'm keen on getting Windows hosting at WebAfrica but their support just isn't as good as it was a couple of years back. Well except if you skip their normal channels and go via MyBroadband. So I just don't know.

 

[Murmaider]

Wow... talk about using Cloudlinux the wrong way...
WA intentions are right, but the execution is bad.

- Firstly they shouldn't be using suphp, its slower and fcgi is better.
- The reason your memory is spiking is because they are defining the Virtual Memory limit in a Cloudlinux 6 install, where this should be set to Zero and only use the Physical memory limit. The Virtual Memory bugs out causes php 500 and 503 errors and should only be used in CL5.
- Their IOlimits are set to low, so instead of allowing the scripts to execute quickly and complete, they are forcing the process to queue and drag out longer than it needs too.