|
just start with docker first, make some images and get used to the tools. kubernetes is a lot easier to install on small hardware than dc/os (it needs like 3-5 vms minimum lol) and anyway i think mesos isn't that popular anyway from my observations. theres like a billion guides out there https://www.digitalocean.com/community/tutorials/how-to-install-and-use-docker-on-ubuntu-16-04
|
#
¿
Jun 8, 2017 04:54
|
|
|
|
| # ¿ May 14, 2024 01:02 |
|
|
also ive used https://hyper.sh/ to host some things like gogs. its pretty sick because the api commands are exactly like deploying/building anything on docker
|
|
#
¿
Jun 8, 2017 04:55
|
|
|
containers and their schedulers aren't anything new. openvz and zones and yarn etc have been around forever. it's just docker made the build and image management part very easy. IRL deployment is still a major pain though unless your app is totally stateless/doesn't need any local persistence
|
|
#
¿
Jun 8, 2017 17:16
|
|
|
like for example doing something as simple as saving images to a disk, in a clustered setup, is not possible. you now need an object store or some distributed storage. it ratchets up complexity a ton. most people don't need to deploy infra like this
|
|
#
¿
Jun 8, 2017 17:19
|
|
|
VOTE YES ON 69 posted:this is actually really easy and idg why people bring it up so much it all falls apart the second you need persistent storage. you're left with managing the distributed garbage you posted or implementation specific block device janitoring that follows the container, like rexray. basically just turns into something you could have done with vms easier
|
|
#
¿
Jun 8, 2017 19:32
|
|
|
which is why yarn/mr was married to a distributed file system by design (hdfs)
|
|
#
¿
Jun 8, 2017 19:34
|
|
|
it's really intended for large scale distributed deployments through a scheduler, running docker on your $5 vm isn't gaining anything useful. you can do all the same build automation with packer/ansible still worthwhile to learn though 
|
|
#
¿
Jun 8, 2017 20:42
|
|
|
yes i agree paying amazon money does make stuff easy
|
|
#
¿
Jun 9, 2017 07:43
|
|
|
cool story though: s3 isnt block storage
|
|
#
¿
Jun 9, 2017 07:44
|
|
|
just put your database and kafka logs on s3 lol no big deal
|
|
#
¿
Jun 9, 2017 07:46
|
|
|
Shinku ABOOKEN posted:i want to make a personal website. what container / cloud should i use ?? digitalocean
|
|
#
¿
Jun 10, 2017 22:59
|
|
|
learn ansible instead
|
|
#
¿
Jun 11, 2017 23:51
|
|
|
it's literally possible to have a git repo with all your playbooks same way you can have a git repo that has all your puppet modules
|
|
#
¿
Jun 12, 2017 04:15
|
|
|
|
| # ¿ May 14, 2024 01:02 |
|
|
and lol at chef and puppet 'scaling' organizationally at all. ive seen zero good deployments in 3 giant corporations. it's always a giant pile of poo poo with ridiculous edge cases. did they use hiera? are they depending on some bizarre facter bootstrap to work? are they using roles or is everything hardcoded in the module? do you need to create ten dozen different node files for it to work? are they using some kind of in house enc? did they use LWRP, or libraries, or HWRP? has everything become a custom rube goldberg machine written directly in ruby?? do you need to refactor this dumpster fire to get anything done? yes
|
|
#
¿
Jun 12, 2017 04:22
|
|