Moving has been hard work as I no longer have a help-desk to get me out of trouble.
The up side is it is about 1/2 the price with better performance.
No more cpanel unfortunately but the free virtualmin isn't so bad.
Cloud servers look promising but the choice is not a simple one.
This rave continues here.
In theory the VPS package is very good but in practice it hasn't met my expectations. A reasonably priced managed VPS is almost ideal for me. It just didn't perform as advertised.
The migration is complete and the VPS is now working well. The search for a new host is off for now.
The load is now fine but there were still problems with the VPS pushing the "shmpages" limit - that's "shared memory pages".
I rebuilt apache without the acceleration option (or frontpage) and shmpages usage went from over 20,000 to 100.
Once again I hope this is the end of the server crashes.
The plans are easily upgradeable if load grows. VPSlink for example have 7 plans from $8 to $260 per month - cheaper for long contracts.
Addsense revenue is close to zero and currently wouldn't even cover the $8. I'd also want cpanel if possible and this is extra.
Anyway we could try a plan for a few months and see how it goes.
I'm having a minor problem with hotlinks to nerdipedia and have been unable to get the bluehost hotlink protection to work - I'd expect better control over the server will fix this.
I've kind of expecting a general move away from shared hosting to some sort of managed VPS as it would be fairer all round.
The reviews for cheap VPS were pretty bad - I'd be worse off than I am now.
We've pretty much decided on JaguarPC as the new hosting company.
The budget managed VPS package is under $20/mth but adding fantastico, cpanel and the memory it needs is an extra $15.
Adsense revenue is practically zero so I/we will have to fund it.
Should decide today what options (OS,panel memory) and sign up.
I contacted Jag pre-sales help and asked about cpanel memory.
I asked three questions (sort of) and got a one line reply five hours later which didn't really answer me.
It looks like Jag help is as useless as the rest.
I know I'm not the first to notice many people don't seem to read entire emails - particularly if it is a mix of quotes and relies (this wasn't).
Newbies can be forgiven but I.T. professionals should know better.
Not off to a good start here.
I signed up for a JAG VPS plan and at the end of the process the web-page said to keep the browser open until the end of their verification call.
That was two hours ago.
It is night here and very early in the US. I've read the sales staff are only there office hours.
So if they call it will either be when I'm asleep or getting ready for work or when I'm at work.
We'll see how it pans out.
Hosting companies seem to go overboard with verification - particularly when root access is involved.
So how long do they expect me to keep the browser open?
They know their customers aren't all in the US timezones - they could handle this a whole lot better.
Emailed them got a response in under two hours so things appear to be in motion.
It is 1am here 8am-ish in US depending on which zone they are in.
Back to bed and fingers crossed VPS is only one sleep away.
I'm having serious signup problems with Jag. My forms are coming through blank.
They think it's a proxy problem - I don't.
I've asked them to process manually and this has been ignored so far.
I can try again from work or I can get someone else to sign up for me.
After 3 days of this I'm getting a little peeved.
They've manually entered my details and it *should* be set up tomorrow.
Another day passes and no account.
I emailed again and my ticket is now being forwarded to management.
If we get through this I hope the service is worth all the drama.
VPS working at last.
Still a few problems to sort out and testing to do.
The new server looks impressive.
Moving a wiki isn't as easy as moving a static html site.
There was a database to import, image and file directories to merge and paths to fix because the new file structure is slightly different.
We're getting occasional server crashes. The system come back automagically.
Down time is maybe five minutes.
I think we (Arjan or I) are causing it somehow.
We use cpanel which is a memory hog and is possibly a bit buggy.
Haven't gotten the .org version of the domain working.
I might be able to do it by editing .htaccess but was hoping there'd be another way.
(later it is easy once you know how by parking the .org on top of .com)
One source of DNS confusion for me is we seem to have a recycled VPS and the previous owner's name servers were still set up. I'd assumed jag had set this up for us but this doesn't seem to be the case.
In the account setup info it says our name servers are ns1.nerdipedia.com and ns2.nerdipedia.com but this isn't the case and explains why I had trouble.
We've used the "funny" servers so I can't simply delete them. We only have one spare IP and I've used this for ns1.nerdipedia.com.
It seems one NS is enough to make things work. After I inform Arjan what I'm doing (after the sun rises) I should be able to delete the old names and fix things.
I couldn't delete those name server entries because the were registered with icann.
Support fixed it fairly promptly.
Just when offline for about 40 minutes.
I couldn't upload files to the wiki.
The permissions for the upload directories had changed - I hadn't touched them.
In trying to fix that the server crashed with internal server errors.
You know "contact the web-master" - this is why I talk to myself so much.
I had no clue and found enough info online to find the logs - and discover the crash was from public-html having too many permissions set.
[Mon May 26 20:10:44 2008] [error] [client 188.8.131.52] SoftException in Application.cpp:246: Directory "/home/nerdiped/public_html" is writeable by group
Again I hadn't touch it.
It is interesting to note the server logs go back to last August.
Support now confirm there was an account setup error.
"The burstable memory was not set correctly that has been done now to 512M"
I think the server problem are finally solved - we will have to wait and see how it goes.
Finally managed to transfer the domains to AletiaNIC.com and register our name servers.
I'm now doing daily backups to my bluehost account.
Wiki traffic and add revenue have been pathetic so we will probably host a few paying accounts on the server to cover costs.
4'th July 2008,
Moved eddiem.com to VPS. I was having trouble with server side trojans injections at bluehost.
The trojan is called HTML.IFrame-10.
Moving to our VPS at least gave me access to an antivirus scanner which showed up some 632 infected files - all html or php "index" files.
The move was slow because I moved a 1/2 gig backup file from server to server with my cable.
"Restoring" the account was easy and almost everything worked.
My account had many redirectors, wikis emials accounts etc.
I could create accounts. It looked like a disk quota problem even though we have plenty of space.
Jag help fixed it - yet another wrong configuration at there end.
Apart from lack of use the VPS is working pretty well. I've implemented fastCGI and PHP acceleration so we should be in good shape to handle increased traffic.
NP is getting around 700 page views a day from 200-300 visitors, traffic is growing slowly.
Eddiem.com still gets more traffic. In total we're only using 16 gig a month bandwidth for all accounts but I use another 15 gig for doing off site account backups. Our plan includes 150 gig a month so we are only using %20 or so.
There have been lots of unauthorised login attempts lately and I've systematically blocked to offending IP blocks - starting with China and Korea.
Most attempts are to non-existent accounts but others go straight for root access.
We have a fairly secure password and anti-brute force features so the chances of a hack through password guessing is unlikely.
The IPs these hack attempts come from are already known on the internet - surely these guys could be found and made to break rocks instead.
Neither Arjan, myself or another Jag help guy could find any evidence the VPS was compromised or hosting badware. There were no new scripts and no increase in bandwidth. It is possible we were hacked and they cleaned up after themselves very well but it is more likely our IP was being spoofed and our server had nothing to do with it.
The alleged attack was on SSH so blocking port-22 should pretty much clear us from blame in a future attack. Not all severs us port-22 but hackers are unlikely to go for anything else.
I would expect jag could trap SSH packets coming from the VPS and this is what should be done in this sort of situation.