Best posts made by stacksofplates

https://hooks.technology/2017/09/30/automated-ansible-testing-with-molecule/

Write up I did for automated testing of Ansible. Just didn't feel like copying it over here.

Seriously?

https://arstechnica.com/information-technology/2017/10/equifax-website-hacked-again-this-time-to-redirect-to-fake-flash-update/

I put Ubuntu back on my laptop and I've been playing around with some of their stuff again since they have some pretty cool tech (Juju, MaaS, LXD, etc). I hate the fact that Red Hat pretty much abandoned LXC. It's painful on Fedora and even worse on CentOS.

I might start using Ubuntu more, and 17.10 is pretty impressive as a workstation OS.

Here's how easy it is to install Rocket.Chat with Snap on LXD.

@scottalanmiller said in Veeam Backup and Replication for ESXi:

@stacksofplates said in Veeam Backup and Replication for ESXi:

@storageninja said in Veeam Backup and Replication for ESXi:

@scottalanmiller said in Veeam Backup and Replication for ESXi:

@storageninja said in Veeam Backup and Replication for ESXi:

@scottalanmiller said in Veeam Backup and Replication for ESXi:

@storageninja said in Veeam Backup and Replication for ESXi:

@tim_g Welcome to Linux. All kinds of shit breaks on new kernels...

If you want a *Nix based desktop that just "@#$@% works" and has a bare metal backup may I suggest MacOS and Time Machine? You can from the BIOS re-install bare metal even from Apple's servers (PXE over WAN).

What's bizarre to me is that you would waste time on forum support for something you complain enough to spend 30 minutes troubleshooting in forums or complaining here about. A license is $50-$30 per desktop per year. You apparently WANT a backup of the desktop and see value in it, but don't want to spend money....

I want a functioning car. I don't want to pay for gas. Sadly this doesn't work.

What crap software are you dealing with? Don’t know any software with this kind of problems. What crap does VMware run that they can’t update?

Not understanding this question....

20 years on Linux and I’ve never seen this issue once. Use enterprise software and you should be good. How would any useful software carry kernel dependencies?

Block level storage hooks can often end up there (although there are exceptions like the micro-redirection in VAIO).... IN this case, it's Veeam making the software (curious where VMware came into this).

The other issue is API's changing that are called. People sometimes change them, sometimes they break etc.

Again really the only time API changes should break anything outside of the OS is when software is using 3rd party kernel modules which is against best practices.

Which in turn makes those apps generally crap. (Zerto, for example.)

I agree 100%. And in the case of Symantec it created a horrible vulnerability where all the user had to do was receive an email (not even open it) and they were infected because of the way the kernel module was written (on Linux).

This is overkill but it’s me trying to run my house like a real system. I have dual CentOS BIND DNS servers and a CentOS DHCP server. The DHCP server is running on KVM, DNS servers are on OpenStack. Ansible updates one DNS server, reboots it, then moves to the other and does the same. It does that every night. DNS and DHCP are great practice for automation. All of my stuff (home and work) uses reservations so I just add a server to my dictionary and it’s created by Ansible in both DNS and DHCP at the same time.

If you own a domain you could do this with CloudFlare.

https://mangolassi.it/topic/13074/dynamic-dns-with-cloudflare

This gets even better. You can follow the responses between the site and the scammers!

This is ridiculous

https://www.techspot.com/news/72612-western-digital-cloud-drives-have-built-backdoor.html

I'll recommend Drupal. It's the most flexible and you can build anything. It's great for very simple sites or really complex ones.

@dashrender said in switch DEs:

I'm at a complete loss.

I installed Fedora 27 Desktop edition with gnome (huge mistake - damn is gnome ugly and near useless)

So I installed LXDE (Don't ask me how I don't recall).

[h@plex-host ~]$ sudo dnf list installed *lxde*
Installed Packages
imsettings-lxde.x86_64                  1.7.3-1.fc27                    @updates
lxde-common.noarch                      0.99.2-3.fc27                   @fedora 
[h@plex-host ~]$ 

[h@plex-host ~]$ dnf group list installed
Last metadata expiration check: 0:21:32 ago on Sat 27 Jan 2018 08:20:53 AM CST.
Installed Environment Groups:
   LXDE Desktop
Installed Groups:
   Administration Tools
   LibreOffice
   GNOME Desktop Environment
   Fonts
   Hardware Support

Upon rebooting, I see the gear icon next to the logon button, but even though I choose LXDE, gnome still is what loads.

Just install the LXDE spin from the iso. I’m curious as to why you say GNOME is ugly but want to install LXDE. It’s one of the worst looking DEs around. LXQt is minimal like LXDE but uses Qt instead of GTK. But using Qt also changes your set of applications that you use (without installing the GTK libraries also). LXQt may have the GTK libraries already installed, I haven’t used it since it was beta, but the environment is made to run Qt applications.

@scottalanmiller said in Curl Issue:

This means that the link that you have is bad, but they are redirecting you, potentially, to a good one. cURL, by default, does not follow redirection. But there is a setting to enable that.

Set: CURLOPT_FOLLOWLOCATION to 1

You can just use -L to follow redirects.

There's a couple ways. You could either create the VM manually and do your setup and then clone it later, or use virt-builder to pull in pre-built images. I usually go for virt-builder. The images are smaller and have things preset for KVM (like console access on ttyS0).

Just as a side note, one cool thing you can do with your templates is inject updated packages. Just do virt-customize --update --selinux-relabel on your template and it will update the packages inside and relabel SELinux (obv only use if it's a RHEL based VM).

I made a simple role to set up GRAV. Sadly, they hard code the version in the URL download so if it changes, it will have to change in the role as well. I could parse the page and look for the current version text, but they have the same text for the beta also. Oh well.

Here's your requirements.yml:

- src: https://gitlab.com/hooksie1/ansible-firewalld.git
  name: firewalld
  scm: git
  version: master
- src: https://gitlab.com/hooksie1/ansible-httpd.git
  name: httpd
  scm: git
  version: master
- src: https://gitlab.com/hooksie1/ansible-grav.git
  name: grav
  scm: git
  version: master

And here's a sample playbook:

---
- name: Set up GRAV
  hosts: grav
  user: vagrant
  become: true

  roles:
    - { role: firewalld, firewall_services: "http" }
    - { role: httpd, www_html_override: "All" }
    - { role: grav }

Just grab the roles with:

ansible-galaxy install -r requirements.yml -p <your role path>

Then run your playbook:

ansible-playbook grav.yml

Go to your URL and you'll have the create user page:

@tim_g said in Linux Network monitoring:

Yeah Grafana is awesome for having things look really pretty. Personally, I'd prefer a more complete solution like Zabbix.

You can do both. There's a connector for Zabbix and Grafana. Here's a demo: http://play.grafana-zabbix.org/dashboard/db/grafana-zabbix-demo?orgId=2

@scottalanmiller said in Storage and Data Locality:

@black3dynamite said in Storage and Data Locality:

@tim_g said in Storage and Data Locality:

@stacksofplates said in Storage and Data Locality:

@tim_g said in Storage and Data Locality:

ReaR is great as agent based backup for image based recovery and inceementals. For file based, I like FWBackups.

ReaR works great on stateless VMs, really easy and quick to recover.

Why are you backing up stateless machines?

Policy and DR.

With stateless machines, wouldn't it be easier to automate the deployment, especially if your data is stored on another virtual disk or storage appliance?

Exactly, that's the theory. Have a system like Salt, Ansible, or just a build script that can restore the box automatically and another script or command that can restore the data, rather than the system. This provides not just lower cost and faster backups, but faster and more flexible restores.

Ya. I just back up GitLab. There’s a couple VMs with data but not many. All of our workstations are stateless. If there’s an issue I just kickstart again and Ansible does the work.

It's definitely not a joke. It's useful for the purposes it was built for. It's hard to manage by itself, that's why you run an orchestration/service discovery with it. You define how your infrastructure is supposed to look in YAML and k8s will build it. A small example would be you've got an RDS instance running for NextCloud and some other services. You tell k8s that you want 3 replicas of the NextCloud server running and to expose on a certain port. K8s then does all of the service discovery and port mapping for you. If a node goes down, it brings up a pod on another node. On AWS and GCP it even automatically works with ELBs and whatever GCP calls their load balancer (I forget). So tomorrow you want to deploy the new updated version of NextCloud. Tell k8s to deploy the latest container version and only destroy after another is created, that's it.

It definitely has it's place. It's made for immutable, distributed services and systems. It does that job well.

The type of snapshot you created is an external snapshot. The type that you create when you use the Virt-Manager gui is an internal. The internal all exist in the .qcow2 image and are copy on write. The external are AOW/ROW so they have to be block committed to the original backing store before you can delet them since all new writes/reads have been directed to the new image.

Just as a side note, if the guest is running a database it's best to install the QEMU guest agent. Then pass --atomic when you do the snapshot. This will quiesce the file system and then unfreeze it when the snapshot is finished.

I think you lose a lot of the advantages of you treat these as just VPS vendors and not as cloud providers. While Vultr does have an API most tools have DO already included as providers. So I think the advantage of being able to use tools like Terraform, Packer, etc natively gives DO an advantage.

Use submodules for this.

git submodule add https://gitlab.com/<user>/repo repo