Oh it does look like Garry is on the ball after-all :)
up_the_irons: How are you guys going for IPv4 addresses? Is it getting tight?
Oh I just worked out garry == up_the_irons
DDevine: IPv4 isn't real tight for me yet, but I'm still acting like it is so I can avoid that situation for as long as possible ;)
ISO changed
Thanks
Some people are *really* determined to bury their head in the sand when it comes ot the reality of v4's mortality.
"We can NAT, we have the technology!"
idiots.
Crap, I ran into this problem in testing. To boot in a virtualised environment after installing you need to eject the media.
up_the_irons: Can you do that for me?
DDevine: huh? the media will not boot by default
Yeah but it has something to do with the way Fedora/RHEL 6  installs GRUB I think.
Counter-intuitive but I've seen it a few times.
DDevine: i actually can't eject the media, but i can make it look like you don't have a cd-rom drive
That will work.
I was just loading up the installer to try to eject the media form there -don't know if that would have persisted through reboot anyway.
up_the_irons: Can you let me know hen you have removed the cd device?
DDevine: done
you need a full shutdown and boot
yep
Hmmm... does a "hard shutdown" from the portal work?
Still thinks it had a CD drive (but no media)
Hmmm maybe I screwed up the installation.
I'll try an install with the net-installer on my local machine and see if I get the same rsult.
1.5MB/sec - I love my ISP.
Yeah I must have screwed up teh installation up_the_irons
ok
Can you re-mount the image?
k
DDevine: done
new website is about to go live!
ooh.
Yep - install exited abnormally. It must have done that last time too.
ouch
i suppose secure link is not necessary ;)
http://www.arpnetworks.com/
up_the_irons: Looks good
Yep looks good. Not too wanky but quite modern all the same.
thanks!
cool
What framework?
sinatra
That's an icky Ruby thing right?
hahaha
yeah
Sinatra rocks; if you don't like it cuz it is a Ruby framework, then you're missing out ;)
There is a selection of nice frameworks for Python and PHP.
ew, PHP
I'm always slightly wary of things like that which are their own webservers. I know Apache (for example) has done lots of work on being scalable, but I'm never sure things like Sinatra and Node.js won't fall over when you push them too hard
I'm moving off PHP, my new site is Django based. Unfortunately I have to keep PHP around for Roundcube and Postfix Admin because there's no suitable non-PHP replacements.
I could probably write a replacement for PostfixAdmin, but sadly not for the webmail.
plett if you put other webservers behind Apache with WSGI then you don't have anything to fear really.
In-fact, that's how you're *meant* to deploy those other non-PHP things.
DDevine: Doesn't that just mean that every hit on the Apache server is translated into a request to the WSGI application, meaning that you now have two things that you need to ensure handle the load?
I may well be very wrong, I've not used WSGI
I'm not entirely sure, but when you put it behind Apache with WSGI the behavior is much more controllable and predictable than using the server that comes with it.
You can also put in static file serving stuff and whatever to deal with the load.
And the usual tricks.
If you design your stuf nicely you can just deploy another node to load-balache with and attach to the central database.
I'm a luddite, I'm much happier pre-generating static files where possible, rather than generating them on the fly
You can do that with these fancy frameworks.
But that's not very Web2.0 of me at all
You can do it with m4 and a Makefile too. I wouldn't recommend it though.
There's no real need when you use a templating engine such as Mako - it will handle all of that stuff for you in the best possible manner.
Tell it to cache where, when and how.
Far out... https://bugzilla.redhat.com/show_bug.cgi?id=618227
This seems to be my problem.
Apparently CentOS 6 will start syncing to external mirrors *very* shortly..
Maybe that will not have the same problem...
http://qaweb.dev.centos.org/qa/blog
I might try installing using SL testing media... Though my damn admin is unconscious.
DDevine: Yeah, we run a Centos mirror, and have been told that it is going to be synced today
Cool
I found that you can launch with the "nokill" command and it stops anaconda from eating itself when things go wrong.
Deinitely need to file this little gem away because it seems to have worked.
I'm so happy I found that fix... The amount of times I've had Anaconda kill an installation because of some silly little detail...
This machine seems snappier than my last.
I just reordered a server from Arp. I have no idea what i was thinking canceling it in the first place. Shared hosting is like living in a really crappy apartment complex where the neighbors all have yappy dogs
Some of them are obviously on some type of antipsychotic medication
lol
my uptime is 48 days, and its my fault.
no problems here :O
arp best bang for the buck
they also deal with hooligans
instead of just ignoring them.
Very happy with arp myself. I've used about 5 others.
Some were OK, but nothing to get excited about and a few were crap.
You get a good community here too.
Yeah, no doubt. I got tired of waiting for Heroku to fix asset pipelining in Rails 3.1. Also, this app is not getting hit a lot. I can stick it on a $20 VPS for now and save a lot of cash
If I outgrow that VPS on the high end, I'm going to have more important problems than a few bucks
Ugh, Ruby.
yeah yeah
I've resigned to generating a SELinux policy to get MySQL working with Postfix on my new server.
I didn't have to do this on my test server and I can't find what is different between the two.
i know it's a pain, but isn't that sort of the point of SELinux? not the pain, but limiting how apps can be interacted with
well, maybe the pain too
Yeah it is. I'm not goign to just turn SELinux off.
But I went to the trouble of setting up a test server to avoid having to deal with this stuff now.
true... that does seem very strange
did you install any of the software at OS install time on your test? maybe it set up some SELinux routes for you automatically...
I did have to install different because it was requested that I use a smaller install image, but that shouldn't have made a difference.
Because I installed the same packages and everything.
I turned on the same booleans.
the only thing i can even think of is maybe there's some SELinux config stuff that's missing from the smaller image, seems like what you did should have worked
I checked the policy packages, I have the same ones and same versions.
centos?
Scientific Linux
6
Had a bit of trouble generating the policy - I was grepping the audit.log and the policy looked correct but apparently it wasnt so I used ausearch instead.
Not sure what it did differently but it worked.