How often does your coldfusion site go down?

[minli98]

Hi all,

For a few years now, we have been using hostmysite as our web hosting service. I have generally been happy with their tech support, but lately I notice that our site (a shared account) tends to go down a lot (about twice a day). Most of the time, it's just a matter of restarting the server, a process that takes up to 3 or 4 minutes. Unfortunately, the times that the site goes down are often when the site is at the busiest. We have gotten a couple calls from customers who told us about it.

I have requested a switch to a different server, and for about a month or two the downtime seems to have come down, but lately the down frequency seems to have gone up again.

My question is is this normal with coldfusion in a shared environment? If you say that your site rarely goes down, could you qualify that with how often you check for it or if you have a systemic periodic check.

Thanks, look forward to hearing your experience.

Regards,
Min

 

[TamedTech]

I'm sat on a couple of different hosts at the moment doing my CF stuff, I find I have very little downtime, so little that i barely ever notice it.

2 Times a day is deffinatly not acceptable, last time i had a host droppping my site that often we ended having to threaten leagal claims to cover the loss of business.

If your server is dropping offline that often i would suggest that it has serious issues and needs a re-build.

I would just change provider all together and have done with all the bother, i've not used hostmysite but i've heard that there prices are very "competitive" which in my book is a direct translation to "our tech support is rubbish".

I'd be upset to see my host having unplanned downtime 2 times a month, let alone twice a day.

Rob

 

[Wullie]

Most of the time, it's just a matter of restarting the server, a process that takes up to 3 or 4 minutes.

A reboot can stop problems in a lot of cases, but many times it does what you are seeing and it just delays them from happening again. If they are rebooting twice a day, then they either don't know the software on the server very well so can't debug and correct the problems or they are highly overloaded servers.

Hope this helps

Wullie

Fresh Look - Quality Coldfusion 7/Windows Hosting

The pessimist complains about the wind. The optimist expects it to change. The leader adjusts the sails. - John Maxwell

 

[minli98]

Thanks both of you.

It's a bit of relief to hear that having the site go down twice a day is abnormal. I will monitor my site more closely now and decide in the next week or so if I need to be looking for a new host. I just implemented a task that sends an http request to the site every 15 minutes to check if it's okay.

Regards,
Min

 

[imstillatwork]

crystaltech.com 100% best I've ever used.

Kevin

http://www.lot-o-nothin.com/cfml/

 

[pigsie]

When you say shared account do you mean you are on a shared server? If so then it is possible that other sites on the same server are bringing your server down - if possible I would look at getting a dedicated server (especially if your business depends on it).

 

[minli98]

Hi,

For the last 4 days or so, I have been running scheduled task to check our site every 5 minutes. During this period, 7 times the http request sent failed. 2 out of these 7 times, the outtage last for at least 5 minutes (i.e. at the next scheduled http request, the site is still down).

Sounds like a bad statistic.

pigsie: yes, it's a shared server. I have given thought to upgrading to a server with fewer users, but I have been told that even with cheap shared plans the site should not go down twice a day. So, if our site goes down this much, either they didn't do a good job configuring their servers or they are not tracking the users with problem codes.

What do you all think?

Regards,
Min

 

[TamedTech]

I think you should just change host, there are so many out there to choose from, I hop around every now and then dependant on your requirements.

After some good advice from Kevin I've recently transfered my CF stuff over to Crystal Tech and I've been very impressed with the support and features available from my control pannel.

For instants, last night i mailed then at about 11pm with a query I had, they had replied to my query with a solution within 9 minutes, i have never seen support that fast, further-more, thier average response time for the last 8 hours has been 14 minutes!

I can strongly reccomend them as a host.

Rob

 

[minli98]

Rob,

I have heard a lot of good things about CrystalTech and am interested in trying it out. Great support is a must, but right now I am most concerned about down time. Have you ever checked how often your site go down at CT?

I am not looking forward to moving host because I have to move the database (MS SQL), the SSL, the domain, the emails and I need to check to make sure every page works. If I do, I'll go about it slowly.

Regards,
Min

 

[GUJUm0deL]

Just out of curiousity, does your current hosting company have the SQL Server, IIS and CF Server all on one machine? If so, that can one of the reasons why your site keeps crashing. SQL Server is a huge memory hog, and at times can use up all of the physical memory, forcing the site to crawl and even crash.

Another reason can be the queries being run off the site, check the queries and see which ones is a resource consumer.

Try this code on your site for a few days:

<cfset currentpage = "#SCRIPT_NAME#">
<cfset nowdate = "#CreateODBCDateTime(Now())#">
<cfset yourdomain = "domain.com">

<cftry>
	<cfif isDefined('http_user_agent')>
		<cfset browsertype = http_user_agent>
	<cfelse>
		<cfset browsertype = 'unknown'>
	</cfif>

	<cfquery datasource="#default_ds#" name="addnewpage">
		INSERT INTO pagetracker(pagename,dateloaded,browsertype)
		VALUES ('#currentpage#',#nowdate#,'#browsertype#')
	</cfquery>

	<cfcatch type="any">
		<!--- used for error processing --->
		<cfset sendto = "yourName@#yourdomain#">
		<cfset sendfrom = "SMTP@#yourdomain#">
		<cfset sendtitle = "problem with pagetracker">
		<cfset sendtext = "Current Page: #currentpage# and Date: #nowdate#">

		<cfmail	from="#trim(sendfrom)#" to="#trim(sendto)#" subject="#trim(sendtitle)#" type="html">
		  <cfmailparam name="Content-Transfer-Encoding" value="8bit">
		  <cfmailparam name="Mailer" value="#yourdomain#">
		  <cfmailparam name="Message-id" value="<#CreateUUID()#@#yourdomain#>">
		  	  sendtext - #trim(sendtext)#<br><br>
			  The error was:
			  <cfdump var="#cfcatch#">
			  <p>&nbsp;</p>
			  cgi vars:
			  <cfdump var="#cgi#">			  
		</cfmail>
	</cfcatch>
</cftry>

This a pagetracker code. Place this in your application.cfm and when your site goes down look at what page it dies on. If its the same page or majority of the times its the same page you can pinpoint the issue further.

I had this happen at my last job, the reason was a combination of things:
[ol]
[li]The boss had the SQL Server, IIS, Mail Server, CF Server all one ONE box.[/li]
[li]The tables weren't indexed properly[/li]
[li]Queries were badly written[/li]
[/ol]

____________________________________
Just Imagine.

http://www.imajinarts.com

 

[minli98]

GUJUm0deL,

I know for a fact that the SQL server is in a different machine. Don't know whether the IIS and CF servers are together.

One question about your code: if the CF server dies or is being rebooted, then the code would never get executed, would it? So then won't get the email notifications. In my current tracking, I sent http requests from a different machine.

In any case, my suspicion is that our site isn't the reason the server goes down so much (remember, we are in a shared environment). Our processes are not resource intensive. Either there is a problem site that they couldn't identify or the server is just overcrowded.

Regards,
Min

 

[GUJUm0deL]

minli98, even if the email won't be generated because the server is down, it will log in the dB before it goes down. You can look in the table for the last time and see what page caused the issue. This is just another troubleshooting technique. Is your site traffic heavy? If so, run this code for only a few days, if not run it for a week.

The real reason for this test is to check for patterns. If the site goes down around a specific page, or time then you can pinpoint the problem. Specific time doesn't have to be every day at 10AM or 1PM, it can be every 6 hours, or every week around same timespan (like 1-2 PM). If you know that then you can find out if somoene on the shared enviornment is running an cfschedule item that might take up resources, et al. Know what I mean?

Since you know the SQL server is on a different machine and your on a shared enviornment then the real resource hog might be another site. Have the hosting company look into it, if they don't respond fast enough, look for a new host (and if possible have a dedicated server).

____________________________________
Just Imagine.

http://www.imajinarts.com

 

[TamedTech]

minli98 if you are looking at transfering hosts, i would reccomend first setting an ip only type account with the new host, then dupliate all your files, databases and mails over to them.

You can then test everything with them with the IP, leaving your current domain up and running as it is.

Once you are satisfaied everything is running fine then you can just ammend your Name Server addess's on your domain DNS and you'll only experience minimum amounts of downtime during the transfer whilst the new DNS propogates around the net.

As for the up time of Crystal Tech, i've not been with them long enough to say iether way, however i have it on good authority from web-developer friends that it is excelent.

Rob

 

[imstillatwork]

Crystaltech does have downtime. they send you an email way ahead of time and let you know of upgrades or maintenance that they will be doing. On shared servers, it is possible to get a bad query or dumb loop from another customer to crash the database or CFML server. email, database, and CFMX / IIS are all on seperate servers. a bad query or bad cf code is usually no more than one minute downtime for the services to restart (if you use windows hosting, I do) at the worst, I've seen a few minutes for a reboot, but this is very very rare. I have never came along and found my site dead.

as with ANY shared hosting (ANY site for that matter) a little code to check if the database is accessible is a good idea, so you can send a nice message to your visitors, rather than a error message.

I don't have anything major going on, but do get 10s of thouands of individual visitors per month, and don't get complaints of service.

download / pageload speeds are quick too, always.

Kevin

http://www.lot-o-nothin.com/cfml/

 

[minli98]

Thank you all for your advice. I opened an IP-only account with CrystalTech to try it out. What I've found out is while their up-time is better than my current host (HostMySite), it was still not so great. Yesterday, I measured CT's up time at around 97% vs HostMySite's 94% (1% of a day is about 14 minutes). During a 24 minute period in the afternoon, the CT's site was down for 16 minutes. Today is a lot better, with almost no down-time for both of them.

I spoke with HostMySite people and they suggested that I moved up to their premium plan, where the site will be in a server that's limited to 25 accounts (normal shared server has 150-200 sites). I am going to give this a try, hopefully my downtime goes down significantly.