|
leper khan posted:I thought it was still where new c# features came from. It’s definitely this, whatever else it is. F# 5’s focus is to add features to make it a good data processing language, but I have no idea whether they’re responding to a market that already exists or trying to create one. (I know they’ve been trying for a while, but I don’t know whether it worked.)
|
# ? Mar 25, 2020 17:36 |
|
|
# ? May 10, 2024 12:29 |
|
Carbon dioxide posted:"Suitable hardware"? https://www.youtube.com/watch?v=msX4oAXpvUE&t=8s
|
# ? Mar 26, 2020 19:42 |
|
|
# ? Mar 27, 2020 19:12 |
|
What's the right way to commission a new server? My process for personal projects is ssh into the box and tinker with various config and restart services until it all hangs together. I'd like to start doing it the right way.
|
# ? Mar 30, 2020 17:35 |
|
toadoftoadhall posted:What's the right way to commission a new server? My process for personal projects is ssh into the box and tinker with various config and restart services until it all hangs together. I'd like to start doing it the right way. Ansible, Terraform, and Chef are all tools that give you some ability to do this with their DSLs. My team would check configuration into a repo and run on tagging a release tag. I favored Terraform for deploying Openstack servers but each tool has its benefits and detractors. And of course there are more remote configuration tools out there than the ones I've named. Smugworth fucked around with this message at 18:08 on Mar 30, 2020 |
# ? Mar 30, 2020 18:04 |
|
toadoftoadhall posted:What's the right way to commission a new server? My process for personal projects is ssh into the box and tinker with various config and restart services until it all hangs together. I'd like to start doing it the right way. I had to do this for a new Test and Production web server, so I didn't need something that worked at scale, and eventually landed on Packer to fetch an Ubuntu ISO, a preseed file to get that ISO installed into a VM, and Docker Compose to bring up services on the VM. It works fine, so far.
|
# ? Mar 30, 2020 21:09 |
|
I'm partial to Ansible for that stuff, though I've mainly just used it on hobby projects like an 8 Raspberry Pi K8s cluster and a web host VPC thing. Meanwhile I thought Terraform was more about getting the systems up and running? In other words you'd use Terraform to talk to AWS (or equivalent) to create the machine instances, and then Ansible to install your software/configs into those instances. Or has Terraform started creeping into the latter functionality as well? I guess you could build your own AMIs (or equivalent) to run on the instances. I've been mostly successful in avoiding that side of things outside of my own hobby projects. I had previously used Puppet for this, but it was a total PITA because it wants to model everything as a DAG, where arbitrary steps can run in parallel unless you explicitly declare all dependencies between them. As a result it was surprisingly difficult to get dependencies to behave consistently, since all dependency links must be perfectly declared across all nodes in the DAG or else things inevitably end up running out of order in surprising ways. It's also hard to diagnose these dependency bugs without wiping everything on the target, since the initial run will hit problems due to missed dependencies, and then the second run will end up working fine since those missing dependencies will often have been mostly installed by the first run. The whole DAG thing in Puppet feels like they decided to do something over-complicated to look "smart" that just made it brittle and difficult to use with little real benefit. Meanwhile Ansible just structures things as lists of lists, where the lists run sequentially and dependencies are implicitly covered by list ordering. As a result it's much easier in practice to get something that works consistently from the start. And as a side benefit the structure means that if you can express something as a series of shell commands then it can pretty much be directly transcribed into an Ansible script. This could mean that it takes longer to run than an equivalent Puppet script, but I think that by the time you got your Puppet script actually working correctly it would end up looking like a list of lists anyway. I found that this difference resulted in taking an hour or two to write something from scratch in Ansible that would take a couple days to get working consistently in Puppet. Not to mention the additional maintenance load as steps are added over time.
|
# ? Mar 30, 2020 22:46 |
|
Progressive JPEG posted:I'm partial to Ansible for that stuff, though I've mainly just used it on hobby projects like an 8 Raspberry Pi K8s cluster and a web host VPC thing. Meanwhile I thought Terraform was more about getting the systems up and running? In other words you'd use Terraform to talk to AWS (or equivalent) to create the machine instances, and then Ansible to install your software/configs into those instances. Or has Terraform started creeping into the latter functionality as well? I guess you could build your own AMIs (or equivalent) to run on the instances. I've been mostly successful in avoiding that side of things outside of my own hobby projects. This is why when VMs have to be part of the equation, I always gravitate toward immutable infrastructure with Packer. Basically take the same approach as containers, just slower and bigger. My experiences with Ansible, chef and puppet have all been terrible. Especially chef.
|
# ? Mar 30, 2020 23:39 |
|
My experience with management tools is that they usually just introduce some other type of problems. Would rather just deal with regular SSH than some idiosyncrasies from some new platform. That Packer looks real nice though, it'd be nice to set up a copy of the prod VM on a local machine for testing...
|
# ? Mar 30, 2020 23:55 |
|
Protocol7 posted:My experience with management tools is that they usually just introduce some other type of problems. Would rather just deal with regular SSH than some idiosyncrasies from some new platform. We use Packer pretty extensively in my current job to build AMIs for new deployments and thus far I've got zero complaints with it. Especially for the use-case you're describing you could look into using Vagrant (another HashiCorp tool) and the vagrant-specific post-provisioner for Packer. I like Ansible a lot for setting up and managing environments - precisely because at its heart it is just SSHing onto a host and running some scripts. Depending on how strict your IaC requirements are it can sometimes just be easier to use Ansible to deploy your instances, generate the inventory and then run your provisioning plays against it. whats for dinner fucked around with this message at 14:12 on Mar 31, 2020 |
# ? Mar 31, 2020 03:27 |
|
I’m definitely going to check those out. Not having trouble with my current setup by any means, but it could make my life easier and I’ve got a lot more free time...
|
# ? Mar 31, 2020 03:57 |
|
Progressive JPEG posted:I'm partial to Ansible for that stuff, though I've mainly just used it on hobby projects like an 8 Raspberry Pi K8s cluster and a web host VPC thing. Meanwhile I thought Terraform was more about getting the systems up and running? In other words you'd use Terraform to talk to AWS (or equivalent) to create the machine instances, and then Ansible to install your software/configs into those instances. You just reminded me of the smelly things we did with TF to get our machines running once TF provisioned them. TF was great for getting our resources from Openstack, don't know if TF 12 is much better at configuration after you acquire them. I'd reckon Ansible would be an ok approach. Speaking of Terraform, I once uncommented out the configuration that disables its ability to destroy your instances if it thinks it has to, then forgot to comment it back out. Fast forward 3 months and another team member wasn't paying attention to the plan output of TF and it nuked someone's cluster. Sorry customer, there went your data...
|
# ? Mar 31, 2020 18:48 |
|
Smugworth posted:You just reminded me of the smelly things we did with TF to get our machines running once TF provisioned them. TF was great for getting our resources from Openstack, don't know if TF 12 is much better at configuration after you acquire them. I'd reckon Ansible would be an ok approach. Do you use TF to provision your databases? Seems risky!
|
# ? Mar 31, 2020 19:41 |
|
Pedestrian Xing posted:Do you use TF to provision your databases? Seems risky! Only if you consider Kafka a database, which you shouldn't, despite the marketing
|
# ? Mar 31, 2020 22:43 |
|
how guilty should I feel about playing on my switch during meetings when I'm working from home?
|
# ? Mar 31, 2020 22:56 |
|
Are negative levels of guilt allowed?
|
# ? Mar 31, 2020 22:59 |
|
You must be new to remote work
|
# ? Mar 31, 2020 23:04 |
|
Faith For Two posted:how guilty should I feel about playing on my switch during meetings when I'm working from home? You should feel guilty if you don't.
|
# ? Mar 31, 2020 23:47 |
|
My rule of thumb is that if you can tune out the meeting and not have it affect your job in any meaningful way then it’s an unnecessary meeting. Switch away.
|
# ? Apr 1, 2020 00:52 |
|
Don't feel guilty about not being "on task." Consider feeling guilty if, as a result, you don't do your actual job well enough.
|
# ? Apr 1, 2020 01:11 |
|
There is no such thing as guilt in capitalism.
|
# ? Apr 1, 2020 01:17 |
|
what game
|
# ? Apr 1, 2020 04:40 |
|
Faith For Two posted:how guilty should I feel about playing on my switch during meetings when I'm working from home?
|
# ? Apr 1, 2020 12:45 |
|
Anybody here work at a company that went all in on SAFe? We've started implementing it (e.g. doing the bigass whole-day company-wide sprint scoping, planning, story pointing, etc. meetings), and my impression of it is very negative. To the point that I'm not confident in the direction of the company now So many loving meetings now, it's a wonder we get anything done. Is there a way to weather it, or should I consider more than I was before?
|
# ? Apr 1, 2020 18:01 |
|
Pollyanna posted:Anybody here work at a company that went all in on SAFe? We've started implementing it (e.g. doing the bigass whole-day company-wide sprint scoping, planning, story pointing, etc. meetings), and my impression of it is very negative. To the point that I'm not confident in the direction of the company now So many loving meetings now, it's a wonder we get anything done.
|
# ? Apr 1, 2020 18:08 |
|
More than happy to. But maybe not yet
|
# ? Apr 1, 2020 18:17 |
|
Pollyanna posted:Anybody here work at a company that went all in on SAFe? We've started implementing it (e.g. doing the bigass whole-day company-wide sprint scoping, planning, story pointing, etc. meetings), and my impression of it is very negative. To the point that I'm not confident in the direction of the company now So many loving meetings now, it's a wonder we get anything done. SAFe is Agile - agility. It wraps Agile up in a whole bunch of superfluous nonsense ceremony for seemingly no good reason. You're right to be skeptical.
|
# ? Apr 1, 2020 21:14 |
|
Just from the SAFe powerpoint slides I've laughed at on the internet I would run as far away as possible.
|
# ? Apr 1, 2020 21:34 |
|
I love how Agile has four very simple values at the very top of their manifesto, and SAFe manages to gently caress up each one of them.
|
# ? Apr 1, 2020 21:46 |
|
The DOD wrote a nice screed on SAFe called detecting agile BS https://media.defense.gov/2018/Oct/09/2002049591/-1/-1/0/DIB_DETECTING_AGILE_BS_2018.10.05.PDF
|
# ? Apr 2, 2020 01:24 |
|
It is possible to shelter a sane working environment within an organization that has been contaminated by SAFe, but it's difficult.
|
# ? Apr 2, 2020 01:30 |
|
Blessed be the manager who does but I fear for their sanity because it essentially doubles their workload.
|
# ? Apr 2, 2020 01:33 |
|
SAFe is horrible but I miss when our PI planning was done at a banquet hall and there was a huge cart of meat and cheese in the afternoons. PI planning was such a colossal waste of time for devs, but maybe it was because my company at the time did it over the course of 2-3 days.
|
# ? Apr 2, 2020 01:34 |
|
leper khan posted:The DOD wrote a nice screed on SAFe called detecting agile BS I disseminated this throughout my old job which used SAFe just before leaving and I’m not apologizing for it. That company was horrible about mushroom management. The only valuable thing that came out of it was having all the SMEs in one place, and available to chat, during PI planning. Otherwise they still had fixed release dates and the whole waterfall shebang.
|
# ? Apr 2, 2020 01:41 |
|
I'll read that PDF and see if I can't start agitating from within.
|
# ? Apr 2, 2020 01:59 |
|
SAFe is the strawman bad project management approach that you compare the thing you're pushing to in order to make your thing look good.
|
# ? Apr 2, 2020 02:57 |
|
spending 2 minutes looking at SAFe makes me think it was a parody website and some MBA saw it and wouldnt stop trying to send them money and now its real
|
# ? Apr 2, 2020 03:22 |
|
https://www.mountaingoatsoftware.com/blog/introducing-the-lafable-process-for-scaling-agile
|
# ? Apr 2, 2020 12:39 |
|
The offshore guys I work with struggled to get any decent image segmentation completed in weeks. I created a model much more accurate than theirs in one week. Now they’re asking for all of my code... yeah, no, I don’t think so. I’ll point you to the same examples and reading materials I used, but I’m not handing over all my work on a golden platter.
|
# ? Apr 2, 2020 14:36 |
|
|
# ? May 10, 2024 12:29 |
|
Protocol7 posted:The offshore guys I work with struggled to get any decent image segmentation completed in weeks. Are you not working together on something?
|
# ? Apr 2, 2020 15:30 |