Surely some of the neckbeards posting CP anime papes on here must have a professional GNU/Linus admin job, right?
What are your thoughts on managing 100+ Ubuntu servers spread across the US? The need would be primarily for maintaining and updating Tomcat applications. Would something like OpenVPN and Puppet work? Has anyone else done something like this?
100 keyboards and 100 instances of SSH
>Has anyone else done something like this?
No, and you don't either so stop trying to make us feel bad for being jobless losers.
never tried something like this with 100+, but you should be able to pipe a script to the separate servers via SSH.
that's my 2 cents
Masterless puppet is what you want. Maybe VPN if they're sharing servers but try to limit horizontal exploitation as much as possible.
what a good approach is to insult people you are asking help to.
go fuck yourself
where do you think you are?
insulting faggots on the internet is the only way to gain answers
No I am not giving you advice on how to create a botnet c&c server.
Hey where did you get that photo?
>100+ Ubuntu servers spread across the US?
Simple, ssh.
If he wanted to do that he would have asked on Hack(skid)Forums and not on Sup Forums
No idea. Probably somewhere in the depths of chan.
How similar are these 100+ machines? If they're nearly identical try this:
They all serve the same purpose. All mirrored. Thanks I'll check it out.
Seconding this.
I manage 1500 servers with Puppet, have pretty advanced setup. AMA I guess
Ya, this. Or chef, salt, ansible, whatever.
The dumb ass already mentioned puppet so you know he's already sound in mind.
Thanks,
The 100+ Ubuntu servers will all have the same function and configuration. Each will be behind its own router. So if you have any suggestions for WAN system management that would be helpful. I'm not sure what kind of WAN integration there is with Puppet at the moment as we're still deciding what to go with.
I've been considering each. Wondering what will be the best for this use.
Let me know if you guys need more details. Thanks.
Another vouch for puppet, we use it at our 3000 host network and it makes life easy
I wouldn't recommend this if you can't restrict access by IP, but you could just open your Puppet server to the internet. The agent on each node initiates the connection to the Puppet server.
Since your servers all are identical, Puppet might be overkill and something simpler like Ansible might be better. With Ansible it's all over SSH so the server initiates the connection to the nodes. This is the opposite of Chef and Puppet which have an agent that connects to the main server.
pls don't bully our autists.
also freetrial of landscape, which is built for this, on this OS
>Ubuntu servers
>not using VMware tools to manage servers
Are you a hobo fucktard?
Was it helpful?
linux admin here.
>60 ubuntu boxes
>most servers running nginx/apache
>monitor using nagios
first month is cancer, but next days are very chill. since most servers are similar in services, most of the tasks like installations and configs are done by a script.
I seldom SSH to these servers unless I want to pretend like I am working.
It's not any comparison, but Rizon uses Ansible for deployment and it's worked quite well, and they have some weird requirements (installation to home menu, passing variables for making C:line configuration, etc.).
>nagios
Didn't that die off any everyone flocked to some fork of it called Icinga or whatever?
Do you know Python?
If so, then use Saltstack.
Do you know Ruby?
If so, then use Chef.
Don't use Puppet.
Ansible's great for 10-20 servers. 100+ would make it slow.
Don't SSH into each one of them or develop a bash script to do that. That's stupid.
Ansible or saltstack.
Chef is ruby cancer.
> Used to manage ~70 linux aws boxes
> saltstack
> would have preferred ansible because it's easier to set up on a fresh instances
Great, another unqualified user got the job instead of me.
>Ruby
>cancer
I see you hate fun when programming. Ruby is fucking great.
VM and docker
> nagios
> not a superior Zabbix
ISHYGDDT.
Ruby's great for devs to have fun with
Ops guys are usually more familiar with Python though, so a config management tool that uses Python (Saltstack being obviously the best) is preferable
Comfig management code is pretty
Basic requiring no more than if and case statements in most cases, unless you're writing your own resources. If your ops guys can't learn this then you've got bigger problems. I personally like Chef the best since you can write regular Ruby code anywhere instead of Puppets bastardized attempt at a fully declarative language where you can't query OS state except using the exec provider or a custom fact.
> relational db required for monitoring
You and zabbix need to die.
I agree.
I've been using chef to patch windows machines as a makeshift SCCM and it's worked great but it feels slow and clunky due to all the cookbook shit.
It's also a resource hog at the server level, it'll eat whatever ram you throw at it. I love ruby but i'm learning puppet and i'd also suggest puppet.
It's more sysadmin friendly, and chef while a great tool is definitely going to benefit devops guys more imo but i don't manage enough servers to warrant any use.
I'm pulling chef from the 100 workstations have and replacing with puppet soon which runs in half the time and requires way less server resources.
Also I get sick of typing their meme Knife command.
Fuck off back to facebook
Saltstack is actually the best amongst all others.
