the issue of suspending sites, cpu usage,etc

[tcwm]

We pay for your service,
yet you often 'punish' your users for using too much bandwidth, or whatever, which in your opinion is a breech of contract. if we pay for bandwidth, we have the right to use it. Your goings on about cpu usuage isnt valid or logical, if your servers cant handle what you offer, then dont offer it.
Cpu availability is your problem, not ours and certainly we shouldnt be 'punished' for using a service that promises something it doesnt provide.

if your servers need upgrading, upgrade them. If you cant upgrade them, then change your packages. simple.

We are using what is rightfully, legally ours to use, being we pay you for it. It is your responsibility to give your clients what they pay for because if you dont..then you are in breech of contract by not supplying your customers with what you say you will.

you say this in your terms of service:

users may not initiate the following
a) Use 25% or more of system resources for longer then 90 seconds. There are numerous activities that could cause such problems; these include: CGI scripts, FTP, PHP, HTTP, etc.

boys and girls. the fact is, you are unable to offer what you say you can.
you offer 60gigs of transfer, common sense tells someone...hey thats a lot of cpu usage for that type of account, perhaps we should change the transfer limit.

your hardware and staff need to grow with the business. Hostgator I think has incredible potential, but within a year will be just a memory (ha, no pun intended) if some critical changes arent made

[GatorBrent]

http://forums.hostgator.com/showthread.php?t=2242 we have one of the largest of threads ever on our forum you yourself were even complaining. The second we kick your site from the server everyone's problem was resolved and no one complained further.

We never look at bandwidth usage we only look at cpu and ram. You cannot cause problems on a server get upset about it, then get eve more upset when the problem is solved. In this case it was your site from the looks of it.

[Surf-Hut]

Brent (or anyone), question:

When I upgrade a program or upload a new script or anything along those lines, is there something I can monitor, to fend off a "situation" for HG, my fellow resellers and most importantly (to me) my clients and myself to avoid an unfortunate situation such as these.

I mean I think I'm fairly competent when it comes to the front side of the house, but the backend, well that's why I have HG and a managed account.

I've been a reseller for over a year and haven't done much, but am planning on expanding this year, and these cases of sites getting booted without much explanation as to why, scares the s**t out of me. I'm not doubting the fact that they are causing problems on the servers and should be....well I'm not sure if getting a permanent boot is the answer, but that's not for me to decide.

So basically is there something I can monitor myself, especially after an upgrade, and periodically to give myself the best chance of something like this not happening to me or one of my clients. Preventative measures?

Since this is the suggestion forum, I guess my suggestion would be for you to work with these clients to help find the problem, not just say they need a dedicated server. without identifying the problem. Which seems to be from what I've read your standard response, these clients are in turn left scratching their heads as to why they would all of a sudden need this. I know if it happened to me, I wouldn't be able to afford that. A little understanding and help in a situation like this could go along way in solving/helping some or your customer relations problems, I so frequently read about. Especially when dealing with a long time client, I wouldn't think you'd want to lose a client over a faulty script upgrade or similar problem. Not in a highly competitive market such as yours. Maybe I'm way off base, but that's just my 2 cents.

But I'd really like the preventative measures question answered, please.

[Surf-Hut]

I've been a reseller for over a year and haven't done much, but am planning on expanding this year, and these cases of sites getting booted without much explanation as to why, scares the s**t out of me.

I actually forgot to add a point on this. Right now I only offer hosting to my consulting clients, who I have a personal relationship with. This year among other things I was going to offer hosting to others. Right now, I am really rethinking this, for the simple fact that if there is a problem with a client site, it gets booted by HG, and I have no idea why, well that could be the end of my business before it even gets started, one adverse comment to a startup business could mean success or failure. All my time and investment could be for nothing because HG didin't communicate the problem to me, in turn making me look like a fool to a client. I don't mind dealing with pi**ed off clients, it wouldn't be the first time and certainly won't be the last, I wouldn't think about hosting to the public if I wasn't prepared for that. However I and the other reseller's need to know that we can rely on HG for answers when the time comes. The more posts I read, the more I feel like I'd be left out in the cold, should something happen.

This is really the only problem or forseen problem I see with HG, I love the service, have very little downtime am a happy customer. Just concerned HG won't be there for me, when the need arises.

Comments??

[sonic]

Originally Posted by Surf-Hut

...This is really the only problem or forseen problem I see with HG, I love the service, have very little downtime am a happy customer. Just concerned HG won't be there for me, when the need arises.

Ditto!

sonic

[britbob]

I`d like to back up Surf-Hut on his post.

I`ve been following these `getting kicked` threads and it leaves me very worried. Nothing worse than seeing the ole `contact billing` sign stuck on every site you have! How do you explain that to clients.

Anyway, it might be an idea to show folks who are not so clued up on how to monitor usage problems (cpu, ram etc). It`s very easy for say me to install a script, make a few mistakes on a few variables which causes major problems.

I often site here and wonder, hey `I hope none of my sites are causing too many problems`. All I understand is how to check my disk space, and bandwidth usage.

[YaBeCuz]

Here's a forum posting from a different hosting company as their response to sites receiving an error message "500 internal server error". This is how they initially handle sites that probably fall into the category where HostGator seems to typically just suspend or kick off (from what I've read or interpreted in various forums).

This method seems to certainly be more customer friendly and allow constructive dialog about the nature of the problem and possible resolution paths rather than treating every situation with abrupt punitive action. Also, for we resellers, it ensures our customers don't see a "contact billing" or "account suspended" message on their site or ours...

Originally Posted by an uneaten competitor

As some of you may already know, we have set limits to our web servers. You will receive that error message if one of your scripts is using an unreasonable amount of resources.

What does it mean?

Well, if a script uses more than 150MB of memory, you will see that error message. However, a script can't run for more than 60 seconds. The final limit is connected with the number of the processes you have started - you can't have more than 125 processes running at any given time.

If we allow higher limits, the machine will become overloaded and will go down until we reboot it and find out which account or accounts have caused the problem and then disable the script or scripts in question so that it doesn't happen again.

Then we will need to inform the customer and ask him to check and fix his script and make sure this doesn't happen again.

The first solution is better.

I am sure all of us will agree.

In this way, the customer will contact us and we will ask him fix his script instead of putting the server down, disabling the script or suspending his account.

[tobyhede]

We were just victims of this exact issue. Our entire account (hosting 6 domains) was suspended WITHOUT WARNING due to CPU utilization. The problem was in a single script in one of the domains, which has appeared to hit some sort of memory leak or similar resulting in resource contention.

I don't really have an issue with a script being frozen because of resource issues, but I really have a problem with an ENTIRE account being shut down without warning because of a small problem in a single domain, and I also have an issue with this occuring without notification. As my email is hosted by HG, through another domain than the one with the issue, I couldn't even log in to work out what the problem was.

We are currently evaluating our options with regard to hosting.

(As it is, we suspect the problem is actually with a CRON job we are running, we're not sure as support isn't that helpful ... we're using GET, which apprently is a link to a PERL library ... this is running the CPU up to 60-70% capacity, we're pretty sure the issue is not in our script, which is pinging only 5 sites for RSS, which shouldn't be an issue ... our dev and testing environments don't use GET and we haven't had these problems).

[tobyhede]

I can now confirm that the problem was not in our script at all, but through the use of GET triggered through CRON.

Caveat Emptor: You're account can be suspended at any time even though you may not be doing anything wrong.

[gdwoods]

Originally Posted by tobyhede

I can now confirm that the problem was not in our script at all, but through the use of GET triggered through CRON.

Caveat Emptor: You're account can be suspended at any time even though you may not be doing anything wrong.

Pretty scary stuff.

I just moved to Hostgator and am a reseller with 20+ clients, mostly business sites that depend on me to keep them going with email, etc...

So what's the verdict HG? Can you work with us on suggestions for monitoring usage and preventing problems, or are we pretty much on our own until the plug gets pulled?

[overflow]

yeah i would like to be able to monitor my cpu usage that i am creating and maybe even down to what user so I (the reseller) can make sure i deal with it and the rest of my sites don't suffer.

[markgroen]

I haven't had too much problem with sites getting shut down because I test locally first, but have had it happen, nobody's perfect, and the problem was solved quickly for me by removing the offending script and having support turn the switch back on.

Today though I was creating a new account, and noticed an account was suspended two days ago and didn't get a notice of it. Just happened to be lucky enough to click on the link in Cpanel and there it was. The reason for suspension was "account spam directly resulted in a outblaze RBL against cadillac" so am guessing a contact form on the site was compromised.

Support hasn't actually told me which script was compromised, it would be nice to be able to monitor these things ourselves. Unless you happen to be in Cpanel at the exact right time and also happen to be looking at "Service Status", there isn't any way to tell something is wrong.

Even if there is a red flag, I don't see a way to tell if it's a domain on your account or someone else's???

[LadyLarke]

I haven't had any sites shut down due to cpu usage etc that I know of. However, my server has crashed more and more often of late. And I have been asking HG why and how and who. I never get an answer.

I pay more money that I care to to make sure I have this support on a managed dedicated server. Yes, they fix things when they break, but they don't bother to help educate me to learn to do it myself when I ask how or why. And that is like shooting themselves in the foot. If I knew how to fix things myself I wouldn't have to ask for support.

It gets pretty darn embarassing when clients ask why the server is down, again. And I don't have an answer. The more it happens, the more inclined I am to take my 4+ years of business with HG elsewhere.

[ferret77]

wow I see I am not alone

I have just had a profitable amazon script suspended and anther shared account suspended, because of cpu usage

No explenation beyond that except being told to buy a dedicated server

I asked if I could try to fix the script

no buy a dedicated server

can I cache the script

no buy a dedicated server

Can i get ftp access and mysql access so I can do a final backup and move my sites?

No answer its been 4 hours which isn't long but it sure seems long as all my sites are down, and I earn my living totally online

I was about to move my sites over here after having a problem with the way my current host handles resource usage. After reading this thread, and a couple like it, I think I'll stay where I am. Internal server errors are a heck of a lot better then taking down the sites with no warning and perhaps more importantly, no access to fix it.

Thats like a bus driver cruising down a toll highway with no means of telling his speed but if he exceeds the posted limit the bus is instantly impounded right out from under his butt, leaving both him and his passengers stranded on the side of the road till the first of the next month.

Talk about getting hung out to dry in front of your clients.

I find it hard to even comprehend handling the situation that way with any site, but with resellers? Wow, that blows me away. Might be different if there were tools available to monitor it yourself but there aren't.

I do thank you guys for having the cookies to run an open forum. It shows your a class act. I spent a couple of days reading threads and I like this bunch, good company, good forum. That said .... well good bye.

[twisty]

I for one am glad they immediately suspend like this, and if anything, I wish they were even more strict about sites causing resource problems for other customers. What's good for the many should always come before what's good for the few who are causing these server problems (even unwittingly).

I'm noticing random, severe load/memory spikes on our server (gator29) and am not happy about it because it causes our PHP pages to stop loading altogether, or load very slowly. We run a messageboard and there is nothing more annoying than visitors posting replies that don't go through, and having to redo them again. I have to think that some other customer(s) are doing something wreckless to cause this.

Having said that, I do however think it's a good idea for the host (who has access to the logs) to clearly communicate the problem after the suspension, and work with the customer to resolve the issue...and not leave them hanging wondering what they can do to fix it

But IMO the suspension SHOULD be immediate (unless it's a DDOS attack).

[Serra]

Originally Posted by twisty

I for one am glad they immediately suspend like this, and if anything, I wish they were even more strict about sites causing resource problems for other customers. What's good for the many should always come before what's good for the few who are causing these server problems (even unwittingly).

I agree, if some moron is running an Amazon script and eating up all of the cpu juice, then they SHOULD suspend first and ask questions later.

However, they could be MUCH better at informing offenders that their site has been taken down. Many of us make our livings on these machines and it isn't nice to have a problem because someone decided that we took too much processor resources. (which isn't an automated process, I assume)

I have just had a profitable amazon script suspended and anther shared account suspended, because of cpu usage
No explenation beyond that except being told to buy a dedicated server

No explenation? IT WAS USING TOO MUCH PROCESSOR TIME!!!!! How hard is it to understand that your script is causing hundreds if not thousands of sites to slow down! HG rules are if your script uses more than 90% of the processor for more than 30 seconds your account is suspended. (or something close to that). That means that your profitable script was actually using 25+% of the server, leave only -75% for the other 1000 sites on the box. Yea, your script should be bannded, erase, destoryed, eliminated and tossed out.

I don't feel a bit sorry for you for having your account suspened. It is people like you that CAUSE the problems and CAUSE HG to make all of these silly rules.

I earn my living totally online

That is a joke right? You earn your living totally on line and you would risk everything by using a shared account with scripts that eat processor time? For an additional $150.00 a month you could have a dedicated box and never have to worry how much time you were using. Seems like your business priorities aren't right.

[Serra]

Originally Posted by MichaelA

Is it possible for HG to change the wording on that page to say something to the effect, "this site is temporarily offline for repair and/or maintenance". I have been lucky so far, but with my clients and all the scripts that they and I would install to run their site, it could possibly happen to me and I know I would certaily appreciate seeing something written like this instead.

Actually, the suspended page is in your reseller account under suspended.page/index.html

You can edit it yourself.

[silentcry]

That is a joke right? You earn your living totally on line and you would risk everything by using a shared account with scripts that eat processor time? For an additional $150.00 a month you could have a dedicated box and never have to worry how much time you were using. Seems like your business priorities aren't right.

Serra, this is a point that I haven't seen brought up very often.. Yet is 100% correct..

I don't know how someone who relies on hosting people's websites for a LIVING, would rely on a reseller account. Yes, it costs more to go to a dedicated server, but if someone is serious about this business, than they need to act like it..

Too many times on this forum have i seen people saying how their hosting business is their livelyhood, yet run on a reseller account, and wonder why they're 20+ customers have issues from time to time..

[stonecold]

Very right things like Semi dedicated and resller accounts are just shamefull. If you serious about your business then you get a deciated machine. if your more advance you play the licence fees to cpanel plesk or who ever your choice is and find a datacenter that will allow colo . Like i still have an aging NT4 box that sits at GTE's Centeral Office , Now known as verizon as back in the old days you would go to your ISP to have a server colo'ed. I am going to have to go down there and rip out the guts of it and upgrade it. i am affaird to take it home as they probably would not let me back in. ( Verizon not happy that GTE use to allow colos at the CO. But as long as that machine does not looks it GTE sticker ( why i just dont replace the machine out right ) they machine is safe and sound and I only pay 80 for service and the space which is a deal these days... P.S. the question of why if i have a colo i use host gaotr as I would not trust a NT web box if my life depeneded on it and 7 years ago and still do day needed a terminal server for application that myslef and other clients could access.

I'd like to first of all apologize to any and all gators out there who were hosted on the same server as my old reseller account which had a database that was suspended this past Sunday night due to overuse.

I've been on both sides of this issue, first as a customer of another hosting provider who I left and came to HG because of their frequent server crashes due to overloading of their boxes. (It is very fustrating to know that your web sites aren't up when you are trying to develop a pilot program for a new line of business. )

During the last few weeks I have been parsing a lot of varied data sources while simultaneously developing an application on a reseller level package and serving up a handful of small dynamic (yet profitable) sites. All of these activities were database based with the same offending database used in all three.

Now I find myself the cause of what I imagine are quite a few people's disdain, specificlly the host gator support team and the clients located on panoz.websitewelcome.com. <ducks class="rotten tomato" > Jason</ducks>

Unfortunately I had ABSOLUTELY NO idea that I was the cause of any downtime and/or performance issues on the panoz server. That is until I noticed that a utility I had written had ceased to function, where it was working an hour before hand. I soon found out that a database that the utility was running heavy sql queries on had been disabled.

As mentioned above by Brent I immediately contacted support to see "what had happened to my database". I was then informed that my dBase had been locked due to frequent server crashes as a result of that dBase and the scripts used to manipulate it.

Now I fully agree with the notion that if you are using that much processing power you have no right to expect a shared server to meet your demands. In fact I immediately opted to upgrade to a full dedicated server account (even though my business plan did not call for this step for another four months).

However there seems to me that threre are a few things that hosting providers and hosting consumers can do to balance value for the individual against the welfare of the collective group.

1. For the hosting provider: proactive monitoring and alerts. Had I recieved a warning via email upon the first instance of my server abuse or after several "close call" threshold breeches I would have stopped developing immediately, analyzed the culprit processes, and scaled back operations.

2. For hosting customers: An improved awareness of your actions; particullarly how they might affect others. Perhaps prior to recieving database and or scripting activation on all shared accounts, hosting customers would be required to read a brief document on common causes of overloading with a simple quiz afterwards.


As for other lessons learned from this experience:

1. The ol' cliche of "isolate your production environment from your test environment" is worth repeating a million times. Even though I consider all of my host gator sites in pilot/test mode, once a pilot site begins making money, move it to a stable environment.

2. Never underestimate the power of memory leaks and thousands of inefficient sql queries per minute. If any of you have studied asymptotic complexity in college and have forgotten it, it might not be a bad idea to review algorithim efficiency. Sloppy code is sloppy code. Even while developing a proof of concept application where performance is not critical this is important.

So again I offer anyone affected by my actions my sincerest apologies. I would like to also comment on the outstanding services that host gator provides. I have never heard of another hosting provider that offers so many services and such high quality support for the prices that they charge. They may not be perfect, but they are as close to it as you can find in this industry. I believe many people would find it difficult to find a better value for your hosting budget, and I look forward to a long relationship with Host Gator. -Thankfully for the rest of the people out there on panoz, on my own box(es).

Best Regards,
Jason Ross

[PFunk]

I am not on the box that you were on, but if I was, I would be the first to accept your apology. It takes a big individual to admit something they did caused other people frustration especially when you had no obligation to do so. Thank you for being a 'stand-up and take responsibility' type of person. It's rare in this day and age. May your panoz neighbors be so kind. I also agree that Host Gator may not be perfect, but they are the best I've used and I've been hosting since 1996.

[Serra]

Originally Posted by PFunk

I also agree that Host Gator may not be perfect, but they are the best I've used and I've been hosting since 1996.

The only thing that I can see that HG is doing wrong is not being very proactive about notifing people that problems exist on servers. I've moved to a dedicated box so I now see the reports that HG is reviewing for CPU and memory usage and I can see some problems that I should address in a few scripts.

For example:

Top Process%CPU 13.0/usr/bin/perl -w /usr/bin/GET http://www.website.com/~user/cron/cron.php
This cron job is eating CPU, way too much CPU for what it does. I need to find out why and fix it.

The bad news is that when this account was on the reseller box, I had no idea it was causing a problem. (though not a big one).

Reports like these would be very helpful in debugging problems on reseller boxes. I understand why HG can't just make these public, because they do contain things that shouldn't be public knowledge, but this type of reporting would solve many problems. Also, when a site is closed for CPU or memory usage, they should include that line from above and show that the log shows just how much of a problem the site was.

Also, when a site gets suspended, I agree, it should happen right away without prior notice, but HG should do EVERYTHING in their power to notify the user. Email and phone calls, even an automated phone call, would be awesome. (like those phone calls we get from blockbuster when we keep a video too long).

I don't want to hear from a customer that my site is down.

As for the suspended page, I'm fairly sure that resellers can edit their own suspended pages.

G A T O R B R E N T - R E A D T H I S ! ! ! ! ! !

There was some discussion of a 500 error instead of suspending the site. (in this thread from Feb). Now that HG is running Mod Security, they no longer need to suspend sites for script problems. They can ban the script in Mod Security until the customer contacts them by modifing the mod security on the specific server. This allows the site to basically function, but denies the script the ability to run. That is FAR better than suspending the account and a policy that HG couldn't do before, but should do now.

[JBenedetti]

I too have been on both sides of this issue, and completely respect Hostgator's monitoring of the servers and blocking problem scripts as a means of protecting the other sites on the server. I only re-sell to lightweight sites for friends or really small clients- don't want the headaches associated with being perceived as a host; but I have 15 or so mid-size clients set up on different shared and semi-dedicated plans with Hostgator, so if one of my sites is causing a problem on a shared server, then it could very well be affecting another one of my clients too, along with everyone else on the same box.

One of my clients, on a Swamp plan, had their forum shut down for a couple of days recently, because it was supposedly eating up too many server resources. Hostgator put up an Error 500 page where the forum was, leaving the rest of my client's site intact.

Tim G, an HG support rep, did an excellent job of working through and explaining the problem with me.

BUT
It would be nice to have some sort of heads-up from HG that my client's site had a potentially problematic script. I just woke up one day to dozens of phone messages and emails from my clients and users of the site... telling me that the forum was gone. This wasn't all that great. I certainly don't want one of my sites to be causing anyone any problems, and would monitor this type of thing myself to make sure I wasn't causing excessive server load, as a means of avoiding any problems and waking up to client and customer freak-outs.

SO
Any chance some of us could have a means of monitoring the load our client sites and scripts are putting on the server? Or at least, could we be guaranteed some sort of fair warning so we can correct things before there is an issue?

[Serra]

Originally Posted by JBenedetti

One of my clients, on a Swamp plan, had their forum shut down for a couple of days recently, because it was supposedly eating up too many server resources. Hostgator put up an Error 500 page where the forum was, leaving the rest of my client's site intact.

SO
Any chance some of us could have a means of monitoring the load our client sites and scripts are putting on the server? Or at least, could we be guaranteed some sort of fair warning so we can correct things before there is an issue?

I'm glad to see HG isn't being as heavy handed as they've been in the past. I would be nice to be able to monitor things before they get out of hand, but most of those tools are at the server level and not at the user level.