Peter's Blog

I mentioned last week how I'd signed up for one of the site5 750g hosting packages and also enquired whether my company could use a similar account for remote backup. Since the latter made me aware of the limitations of the 750g account I decided to cancel my personal account under their 60 day guarantee.

They didn't quibble, they asked me to confirm that I had nothing that needed backing up and cancelled the account, refunding my money.

In all the times I've had to deal with them site5 has always given me great service.

I emailed site5 to ask them whether the 750g accounts could be used for my company offsite backups. The answer was no and it directed me to the site5 Resource Usage Policies on a wiki. It is also interesting to see that there is a limit of 25,000 inodes which equates roughly to 25,000 files. To fill 750g the files would need to be 30 megabytes each.

Conclusion: read the smallprint (and/or the wiki).

I couldn't resist the site5 deal for a hosting account with 750g of space for $7.50 a month over two years PLUS (or rather minus) a 13% Halloween discount. That's an awesome amount of backup space, enough to back up the family photo album of 8g 93 times AND the hosting account service includes my backups being backed up.

I bought a new domain name especially to associate with the new account but I don't feel like sharing it. I'm not sure what actual sites I want to host on it, as I say I got it primarily to back everything else up and I'll be spending the next few days writing rsync scripts to do rotating backups and waiting for 8g of family photos to upload at about 400kbps.

Speaking of broadband, the worst possible thing happened yesterday evening while I was waiting for access to the new site5 account: my broadband died. It was down all evening. First time this has happened in a year since I got it. Wife was busy, had 11 month old baby to take care of, what to do? Watched three episodes of Heroes on PVR. Really good if sometimes a bit talky. Unfortunately I was left at episode nine, Hiro (hero geddit?) was six months in the past trying to stop a rather sweet waitress losing the top of her head. Does he manage it? Dunno, I haven't recorded any more episodes and BBC are now a few months down the line. How to fill the gap? Please, NO SPOILERS. I've already spotted that the waitress happened to be learning japanese six months before she had the bad hair day...

To my tremendous relief the broadband was back up this morning. After a year I'm still with pipex, despite my initial hostility it has been so reliable (so far) that I'd rather let it be. Other WoW players moan about disconnections but these rarely effect me (sometimes at about 9am on Saturday mornings but not for months now). Of course I will be reviewing this if I get any more evenings offline ESPECIALLY now I've savoured the cream of my PVR backlog.

This site was offline yesterday while site5 moved the hosting to a server in Texas (yee ha). Everything was offline for a few hours during which time I changed the DNS so the domains (four in all on this host) pointed to the new server on a different IP address.

When the server came up my generic domain which I use for forgettable experiments was ok but the domain pointers (i.e. secondary mappings of domain names to the same ip address as the primary domain name), including petersblog.org were still broken. I tried to kick petersblog.org into life by deleting the domain pointer and creating it again but then I ran into site5's checking against a whois database, presumably to make sure I wasn't trying to hijack someone else's domain name. It seems that this whois database updates comparatively slowly: when I did a whois lookup the records were already up to date. It wouldn't let me create the domain pointer.

Another domain name on a domain pointer started working later in the day so I ought to have been more patient.

This morning I was able to create the domain pointer and hence I am writing this. Fortunately deleting and recreating the domain pointer has not zapped the awstats log history which is recording half a million hits on petersblog.org over the last year.

A third site was running on Ruby on Rails and Radiant and that was broken: it loaded a page that was totally blank. I presumed that the rails on the new site5 server is different or broken so I did something that was on my todo list anyway, move it to the vpslink server which I have as a better Rails platform.

I've had the site5 account for a year, the contract is up and I may be in a position to abandon it or upgrade. Right now they are doing a special offer of 750g for $7.50 a month (yes, that's 750 GIGABYTES) which is almost irresistable. 750g backup space anyone?

Site5 have been ok up to now for basic web hosting and php stuff. Their rails is not totally hassle free (you can get things running if you spend enough time poring through the forums) and I prefer root access for this. It is still hosting a couple of drupal sites including this blog, mainly because they are tied to php4 which I don't want on my vpslink server (less is more).

The vpslink box has apache randomly crashing, presumably through lack of burst memory. I'm not sure I want to commit myself to it long term. I'm tempted again to get a dedicated host and one of these huge bucket site5 accounts for rsync backups while they are still available.

Disclosure: I was in the site5 affiliate program at one time and even earned a few $$$ from it. I'm not actively in it now, although there may be some old affiliate links knocking around here. I'd rather not take their money and be in a better position to criticise them.

I've had a few emails from people wanting to pay to put links here but I haven't taken them up. You will note that this is one of the few sites on the net not plastered with google ads. I like it that way.

I've got a nice Rails development setup going now. Aptana is a very nice IDE, very powerful, very rich. The Rails development aspect is most useful in being able run applications in development mode on my PC. I use the mysql server on the deployment server through an ssh tunnel rather than install a database on the PC.

I have created a subversion repository on my VPSlink server. I have installed the subclipse plugin on Aptana and checked out my application onto the work PC. I work on the application in Aptana, polish it and when it is ready I commit the changes again through Aptana. Then on my server I deploy the code by getting the new version out of subversion. I don't use a simple rsync to deploy and I haven't got into capistrano (the rails deployment tool) going via svn works for me. I might consider putting up something like trac (or a rails based alternative) to give me a development wiki and web browsing of the repository.

I've been thinking about backups and the current plan is to backup the SVN repositorys and mysql dumps from my VPSLink server to my site5 server and vice versa. The VPSLink stuff is more in need of backing up since I am the administrator of that one, the site5 account includes daily backups. I was tempted by rsync.net a nice, simple, flexible and cheap remote backup option but since I already have two accounts with ssh access and 35G of space between them I don't think I need to spend more money. Rsync.net looks appealing but the main thing it is lacking for professional purposes would be Windows file permissions, otherwise I might consider it for backing up the servers at work: being able to recover all ones files is good, not having to spend a week fixing the spaghetti mess of windows access permissions is better.

One major change is that I have reverted from Lighttpd back to Apache on my VPSLink server. The main reason for this was that Lighttpd seems to have some limitations in terms of supporting things like drupal's urls. Lighttpd would be good for simple setups but getting multiple legacy php and rails applications set up is just as troublesome as with Apache so I've gone back to the devil I know.

I've had Apache fail to start with memory errors a few times. This could well be the 500M limit of the VPSLink coming into play. VPSLink is cheap because there is no swap, 500M is my absolute limit. If I start getting memory problems I will have to consider the options:

• I can upgrade the VPSLink to 1G memory but for similar money I could get a cheap dedicated server
• I could get a different VPS account, one that had burst memory (it's the bursts that kill VPSLink) but I do like the performance of VPSLink.
• I could fiddle about tuning the number of apache processes and suchlike but life is too short for that one

This blog is still on the site5 account. The server is being upgraded soon, maybe this will resolve the loading issues it has whenever I try doing some development on it (e.g run 'top' and shriek in horror).

I was finding Rails development on my Site5 account slightly problematic. There were these issues:

• because of the way the shared hosting was set up you generally needed to trawl the site5 forums to find the special site5 specific hacks to do what you wanted. For example, many seem to have tried and failed to install trac so ultimately one has to compromise over what one can do.
• sometimes it was very slow: I'd find it to be sluggish and checking the load average on the server it would be in double figures: over 300 processes fighting for time with certain people hogging the cpu.

I now have a number of web projects on the go (five including this blog) and I fancied going back to dedicated hosting for the total control it would give me. However there was a problem here: dedicated hosting is expensive and I didn't fancy shelling out that much. After much agonosing I decided to return to using a Virtual Private Server (VPS). I chose to go with VPSLink as their prices were reasonable and it seems suited to what I want. They seem to specialise in VPS's and their site has a blog/wiki/forum which gives them a face compared to oneandone, godaddy or any other big name that only provides vps's to cover the market. I now share a server box with fifteen over guys, each with 512Megs of memory to call our own. Each of us has a 'virtual' server with root access and hence total control over what we do.

On my virtual server I chose to install a preconfigured Ubuntu 7 linux/Ruby on Rails/Lighttpd setup. It is very nice, logging into it with ssh it is hard to tell that the server is shared and is running on a different continent. My rails applications run nice and snappy (once they have done their caching).

I haven't used lighttpd before although I was aware of it as a new clean apache wannabe. While apache is getting bloaty and the configuration files have to me always been obtuse, requireing endless try-this-and-see-what-happens attrition, lighttpd's configuration file was immediately crystal clear: here's the url, there's the directory that serves it. Nuff said. The config file in my installation had a commented out Rails setup which worked on my first tweek.

The only problem I have had thus far with lighttpd is that

/etc/init.d/lighttpd stop

doesn't stop all the lighttpd processes and I have to kill the last one explicitly. I haven't looked into a cause and could even script my way around it if I needed to reboot the server that often.

One possible problem with VPSLink is that their systems are set up such that there is effectively no swap space: if the processes on your VPS use all the allocated memory then tough, they crash. Hence I chose a plan with hopefully sufficient memory (512M) and a light web server. I'm also hoping I don't get too many fcgi processes spawn but I'm not going to lose sleep over that. Should any of my sites get that successful that the server is continually crashing I just upgrade to a dedicated server. (I think I just invited everyone to a DDOS party).

At the time of writing this blog is still on site5. IF I ever get around to porting it to mehpisto then I may move it to the new server. The performance of Drupal on site5 is still acceptable.

Researching online backup solutions: really must backup family photos properly. Options so far:

mozy: 2G free or $4.95 a month unlimited. Automated backup and offers file versioning for a month which would be handy. Files accessable from web.

carbonite: a different shade of mozy, horses for courses. Trial account but no long term free account.

xdrive: 5G free which is enough. Creates a virtual drive which you copy files to however you like or run their backup software. Web based access to files from anywhere, file sharing, lots of nice features. What puts me off? It's owned by AOL

flickr: could just upload multiple G of photos to flickr but since they decided to force me to get a yahoo account I haven't been using flickr (yahoo or AOL, I don't want the hassle). Also the upload wouldn't be as totally automatic as mozy or carbonite.

site5: I could cobble something together on my site5 hosting account: I have about 14G free and ssh access so I should be able to do something but it will be a lashup that I would have to nurse along.

Being a dedicated backup solutions, Mozy or Carbonite would almost force me to do backups and do them well. With anything that looked like another hard disk I would end up with yet another disorganised mess. Less flexibility than xdrive could be a good thing. And by avoiding AOL and yahoo I stick it to "the man".

Just found out that my new site5 server has python 2.4.3. This is Good News. Last time I had a site5 server it was running python 2.2 which was getting a bit ancient. 2.3 introduced generators which have been widely adopted. The latest version is 2.5 which is (flame bait) not exactly a must have, >= 2.3 will do.

Another job for my todo list is to see if I can get django or turbogears running through cgi, but I am not optimistic.

Should I get desperate, site5 offer ruby-on-rails hosting but that would involve selling my soul in a manner akin to adopting asp.net

I've been contemplating recently whether to save some money by changing from a dedicated server back to a hosting service, i.e. site5. I managed to break the server by unwisely doing an apt-get update/upgrade which introduced some horrible dependancy problem.

There seemed to me to be a few possible solutions to this:

• hack around in the apt/dpkg files to try to fix it
• wait for someone else to fix it
• compile it all up myself
• start again from scratch

In considering the latter approach I checked out the http://www.site5.com/affiliates/idevaffiliate.php%3Fid=75 site and found a limited offer of an 80% discount (eight oh) for 15G disk space and 1Tb of bandwidth if I signed up there and then: so I signed up there and then. This should save me over £700 over the next two years (seven hundred). Sold.

The 'condition' of the offer is that I have to tell people about it (the 'tell all your friends' special event). Since I don't have any friends I'll tell you.

The site is now back on site5.com and looking good. It 'feels' as fast as the dedicated server but maybe I'm the only one on it so far.

I got an email alert from SiteUptime to say:

Dear Peter Wilkinson,

This is an automated message from SiteUptime.

Alert Type: Site Not Available Result: Failed Time: September 12, 2005 09:09:43 PST HostName: petersblog.org Monitor Name: Peter's Blog Service: http

There was another email half an hour later to say the site was back again (the polling period is half an hour).

Looking on the server itself:

$ uptime
 09:01:45 up 21 days, 11:21,  1 user,  load average: 0.00, 0.00, 0.00

so the server didn't reboot.

Grepping through the apache access log for SiteUptime..

67.30.130.180 - - [12/Sep/2005:15:11:18 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:15:41:20 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:16:11:21 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:16:41:22 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:17:41:24 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:18:11:25 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:18:41:26 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:19:11:27 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"
67.30.130.180 - - [12/Sep/2005:19:41:29 +0100] "HEAD / HTTP/1.0" 200 - "-" "SiteUptime.com"

Missing entry at 17:11.

Looking around this in the apache access log:

80.88.204.40 - - [12/Sep/2005:17:04:09 +0100]
80.88.204.40 - - [12/Sep/2005:17:04:09 +0100]
194.244.83.8 - - [12/Sep/2005:17:41:16 +0100]
194.244.83.8 - - [12/Sep/2005:17:41:17 +0100]

No traffic between 17:04 and 17:41.

Nothing in the error log. None of the other logs show anything suspicious.

I think my conclusion here is that there was a loss of connectivity within oneandone and my server was temporarily disconnected from the internet. Now lets think, if they promise 99% uptime does this mean the server is running or the server is running and connected to the internet?

Anyway, oneandone sent me this email this morning:

Dear Peter Wilkinson,

Please be advised that due to an upgrade of the 1&1 Data Centre, we will need to shut down your server for a short time while technicians perform an internal realignment of hardware.

This move will be made during the night from 18.09.2005 to 19.09.2005, between 11:00 PM and 06:00 AM.

Your data will not be affected by the move. However, as a precautionary measure, we recommend strongly that you first back up your data and server settings.

Best regards,

The 1&1 Team

so apparently they are shuffling things about and could well be the cause.

The alert emails were forwarded to my vodafone email account but the filters on that prevented me from getting notification text messages (until I got a message I didn't know what 'from' address to set the filter to) so I wasn't sent alarming and costly (2x10p!!) text messages.

Conclusion: I shouldn't panic, I haven't run a site monitor before, this kind of thing probably happened all the time I was on site5 and I was blissfully unaware.