|
VOTE YES ON 69 posted:Nah. DevOps is a pretty common term. And bandwagons are pretty common.
|
#
?
Mar 24, 2017 17:06
|
|
|
|
| # ? Apr 19, 2024 07:13 |
|
|
VOTE YES ON 69 posted:Nah. DevOps is a pretty common term. But devops is a cultural thing, not a job title or a department. If a company is hiring a "devops engineer" or has a "devops team", they're doing it wrong and are just throwing buzzwords around. Someone who "does devops" is someone who works in dev or ops at a company that has embraced the devops culture.
|
|
#
?
Mar 24, 2017 17:34
|
|
|
Pollyanna posted:My resume has only barely marginally improved while working here. I can barely claim to have done anything worthwhile, and it sucks. It's great to keep an eye on this stuff and to be motivated to do better, but do bear in mind that one of the most important things you're doing at this stage in your career is just gaining general experience. In other words there's more to your resume than just checking off skills and launching things. quote:That's good to hear. My current workplace has us all silo'd away and teams barely communicate except in the break room where people bitch about other teams and take potshots at certain key figures that gently caress things up (long story). One big reason that Google gives free meals is to get people together in the same room even if they don't necessarily share a purpose, so that people have those opportunities for cross-pollination. quote:We did post-mortems at my last workplace, and ostensibly it was blame-free, but people got disciplined or fired over them anyway. I find the idea hard to trust as a result. Yeah, it's very easy to cherry-pick the appearance of $famous_company's culture without actually adopting that culture. Not to drink the Kool-Aid too much, but I do feel like Google usually does a good job of practicing what it preaches, though. Certainly when things happen counter to the way they're "supposed" to happen, people get up in arms about it. quote:What do you do when a coworker constantly goes dark when remote/leaves after 12 each day, pushes poorly-made solutions that heavily slow down your team, doesn't offer any code review outside of "please alphabetize these things" or "changes I made to my CICD setup means your entire thing is hosed up now, please redo it all", and has managers asking around if anyone has their phone number so they can get in contact with said coworker? I've fortunately not had to deal with such a situation myself, but I suspect it'd go something like this: * Coworkers tell manager that $coworker isn't doing a very good job * Manager tries to identify why $coworker is having trouble and works with them to improve * If they don't improve, $coworker eventually gets fired In peer evals, I would definitely say things like "$coworker has submitted numerous changes that required extensive remediation by other coworkers (see commits here, here, and here, plus followup bug tickets here and here)" and "I often had difficulty getting in touch with $coworker to coordinate our efforts, resulting in frequent blockages and unnecessary or duplicated effort." Then in the confidential feedback I'd probably tell my manager something like "They're impossible to work with. Why did we even hire them in the first place? They're really hurting everyones' quality of life and are more trouble than they're worth." For that matter, that confidential feedback would have been told to the manager in person rather earlier (since perf is only twice a year, that's a long time to suffer with an rear end in a top hat coworker).
|
|
#
?
Mar 24, 2017 17:37
|
|
|
minato posted:I hear Google recently tweaked their system to avoid people gaming it, where they'd release a project that had significant impact, then immediately abandon it in the next perf cycle because maintaining and supporting the old project had no impact. This apparently had led to a culture of tools and libs that were plentiful but not supported anymore. I'm not aware of any tweaks aside from the reiteration of a focus on landing, i.e. the project actually having an impact through use,and not just launch. I've only been at Google a short time, but my experience with perf is pretty contrary to what's been said in this thread. It's better than the single manager option if your manager is bad, but only marginally as a manager still has a massive impact both in determining rating and promo. It's pretty similar to the performance review process at other companies I've been at where you write great things about yourself and others. I've never seen the point of writing honest self assessment for either yourself or others in this type of format as it's very difficult for it to not be held against you later. The normal format for that is some form of one on one.
|
|
#
?
Mar 24, 2017 18:12
|
|
|
Hold on, folks are writing documents that will go up the chain and determine raises, promotions, and general visibility in the company, and the phrase "soul-baring honesty" comes to mind?
|
|
#
?
Mar 24, 2017 18:16
|
|
|
New Yorp New Yorp posted:But devops is a cultural thing, not a job title or a department. If a company is hiring a "devops engineer" or has a "devops team", they're doing it wrong and are just throwing buzzwords around. Someone who "does devops" is someone who works in dev or ops at a company that has embraced the devops culture. My impression is that a devops team is a team that does both development and operations for an application. A devops engineer I would think is someone who does both development and operations work. Am I wrong, do those not exist?
|
|
#
?
Mar 24, 2017 18:28
|
|
|
It does, it's called a software engineer and it's me bitch!!!!!
|
|
#
?
Mar 24, 2017 18:34
|
|
|
Promotions are primarily determined by the number of chat apps you have launched
|
|
#
?
Mar 24, 2017 18:46
|
|
|
JawnV6 posted:Hold on, folks are writing documents that will go up the chain and determine raises, promotions, and general visibility in the company, and the phrase "soul-baring honesty" comes to mind? It totally does. I'm baring my pristine, perfect, flawless soul to everyone so they justly give me more money & power. I don't see a problem with being so honest. Maybe flawed people would.
|
|
#
?
Mar 24, 2017 18:48
|
|
|
Or how many good blog entires you've written! I wrote a sick nasty blog about a feature I worked on that was huge and "industry first" and "disruptive" and "market defining" before I quit at my last company that would have gotten me promoted.
|
|
#
?
Mar 24, 2017 18:48
|
|
|
Jose Valasquez posted:Promotions are primarily determined by the number of chat apps you have launched Yesterday I halfway considered making a Linux CLI based chatapp with cloud pubsub as a joke/troll/learning exercise before realizing that was 90% IRC.
|
|
#
?
Mar 24, 2017 18:50
|
|
|
My exposure to devops at the beginning of my career pretty much cemented it as "the AWS, Jenkins, and Chef guys". Devops to me means everything to do with getting the app running and available to consumers without literally programming it, separated from "actual development". I don't think I'm happy with that meaning.
|
|
#
?
Mar 24, 2017 19:09
|
|
|
mrmcd posted:Yesterday I halfway considered making a Linux CLI based chatapp with cloud pubsub as a joke/troll/learning exercise before realizing that was 90% IRC. I wrote a chat app once that consisted of a webpage and a single file on the server that was the chat history. You'd just go to the web page, choose the name you wanted to present yourself as, and type your message into a text box, then the text would be sent to the server and written to the file, which every client periodically refreshed. It was actually kind of fun to have conversations with my friends on this janky crappy chat program I wrote to teach myself basic Javascript and CGI.
|
|
#
?
Mar 24, 2017 19:18
|
|
|
TooMuchAbstraction posted:I wrote a chat app once that consisted of a webpage and a single file on the server that was the chat history. You'd just go to the web page, choose the name you wanted to present yourself as, and type your message into a text box, then the text would be sent to the server and written to the file, which every client periodically refreshed. My first job out of college I wrote a chat app inside a post-trade allocation and clerking program used by the brokers at the fixed income firm I worked for. There wasn't much reason beyond I was playing around with the jgroups library and it was really easy to do. Writing a janky chatapp no one wants is practically a SWE right of passage at this point.
|
|
#
?
Mar 24, 2017 19:24
|
|
|
Pollyanna posted:Perf is still reviewed and evaluated by management, so it just shifts the onus of reporting and reviewing onto the employee. It doesn't sound all that different from typical management to me. Pollyanna posted:What do you do when a coworker constantly goes dark when remote/leaves after 12 each day, pushes poorly-made solutions that heavily slow down your team, doesn't offer any code review outside of "please alphabetize these things" or "changes I made to my CICD setup means your entire thing is hosed up now, please redo it all", and has managers asking around if anyone has their phone number so they can get in contact with said coworker? You filter them out in the interview, hopefully, and failing that, you have them managed by someone who will notice this more or less immediately. Management at Google isn't really comparable to traditional management. Most managers (especially first tier managers) are still individual contributors in some capacity, and will be code reviewing patches from their reports. If they don't notice that their report is useless within like a week, something has gone very wrong. (it's harder if they're assholes in the "doesn't play well with others" category, AFAIK what will usually happen is that they'll get dinged for it in perf/calibration, and get explicit feedback of "stop being such an rear end in a top hat")
|
|
#
?
Mar 24, 2017 19:49
|
|
|
I think he meant the engineer part of the title, not the devops part. drat this thread move fast today !
|
|
#
?
Mar 24, 2017 21:00
|
|
|
TooMuchAbstraction posted:I've fortunately not had to deal with such a situation myself, but I suspect it'd go something like this: At any point do you, or anybody besides manager(s), actually try to engage the coworker with whom you have a problem?
|
|
#
?
Mar 24, 2017 21:43
|
|
|
waloo posted:Wow. I mean, yes and no. I don't feel like it's really appropriate for me to tell a coworker "you're doing a bad job, what the hell?" Especially I don't think it's appropriate for a struggling coworker to be told they're doing poorly by all of their coworkers. Letting the manager take point on that kind of thing means that a struggling coworker only has to go through one uncomfortable conversation. It can often be hard to tell the difference between someone who's disengaged because of legitimate issues and one that's just an rear end in a top hat, and I prefer to give people the benefit of the doubt when possible. If the team's reasonably clear that the underlying problem is that the coworker's skills aren't up to par, then we can all pitch in to help train them. If the problem is that they're an rear end in a top hat, then honestly I'm not sure what kind of remediation any of us, coworkers or manager, can take. You have to convince the rear end in a top hat to align their goals with your own somehow, and that's not a skill I have. Which is why I let the manager handle it; they have more training on interpersonal relations.
|
|
#
?
Mar 24, 2017 22:22
|
|
|
TooMuchAbstraction posted:I mean, yes and no. I don't feel like it's really appropriate for me to tell a coworker "you're doing a bad job, what the hell?" Especially I don't think it's appropriate for a struggling coworker to be told they're doing poorly by all of their coworkers. Letting the manager take point on that kind of thing means that a struggling coworker only has to go through one uncomfortable conversation. It can often be hard to tell the difference between someone who's disengaged because of legitimate issues and one that's just an rear end in a top hat, and I prefer to give people the benefit of the doubt when possible. My manager summarized his transfer from an Eng to Management ladder as "Going from looking at a piece of code all day and asking myself 'What about this isn't working correctly and why?' to looking at people all day and asking 'What about this person isn't working correctly and why?'" That, and calendar Tetris.
|
|
#
?
Mar 24, 2017 22:36
|
|
|
The comparison you should be making is you versus the idealized version of who you could be in the near future, if you give it a fair shot (i.e. not killing yourself with work). The more time I spend at Facebook the more I appreciate the ways in which its corporate culture is actually super great at growing people into better engineers and leaders. Pollyanna posted:Goddamn that's hardcore. I have to admit, though, that's something I've been missing a bit in my own career. Some reasons this can be difficult to pull off, though: Pollyanna posted:Yeah, I can see myself immediately putting my foot in my mouth if I end up doing a job like this. I mean, I can do it - lord knows I've been outspoken about stuff at my current job - but I can't guarantee not accidentally pissing someone off by doing so. Pollyanna posted:That's good, at least - I just wouldn't want to see L3 and L4's opinions and buy-in being discounted or ignored. Pollyanna posted:FB does have a Boston office, I think...I could apply to FB too. I don't like PHP though. If they do Elixir or Clojure that'd rule! Also, we don't really do PHP. We do Hack, Python, C++, Java, Objective-C, Javascript. I think we have some ML and Haskell enthusiasts, but ultimately the languages we pick are ones that we can invest in for the long-term (this is why for example Scala is not a particularly good fit for the codebase). Hack is a really meaningful difference because we have a unit-testable, fully linted, IDE-supported, autocompleted, generic-collection-enabled, strictly type-checked system in place with a bunch of tooling around deployment and data gathering. It doesn't resemble PHP in any meaningful way except syntax and compatibility, and developing in it feels a great deal more like Java than like PHP. And that's if you even end up writing Hack. Doctor w-rw-rw- fucked around with this message at 23:52 on Mar 24, 2017 |
|
#
?
Mar 24, 2017 23:48
|
|
|
Doctor w-rw-rw- posted:
Re the perf stuff I've done promo a few times now and it's not terrible, but it does suck forcing yourself to write about how great you are if it doesn't come naturally to you.
|
|
#
?
Mar 25, 2017 00:11
|
|
|
Hughlander posted:So it's Friday and i'm going to be Nothing happened. Nobody has said anything. I quietly threw the packet into the garbage. We'll see if it was just an empty-threat scare tactic or if they just forgot because the upper management managed to miss four different flights today. I'll keep you posted.
|
|
#
?
Mar 25, 2017 00:25
|
|
ing this thread like a motherfucker ; I expect updates and not blue balls.
|
VOTE YES ON 69 posted:Nah. DevOps is a pretty common term. So is "devops" the new "ops"?
|
|
#
?
Mar 25, 2017 01:43
|
|
|
Paolomania posted:So is "devops" the new "ops"? At most places that have it as a role it is a 100% ops role. What you'd call a systems administrator 15 years ago is now dev ops at those places. But again I consider anyone who has a role called DevOps to be a 
|
|
#
?
Mar 25, 2017 02:02
|
|
|
Ops is not at all DevOps in my industry.
|
|
#
?
Mar 25, 2017 02:08
|
|
|
Doctor w-rw-rw- posted:I think (but am not sure) that the Boston office is mostly AI/Machine Learning, but perhaps not exclusively so. When I interviewed there last year, they said the Boston office had a lot of Facebook's version of SRE, security people, and some of the group that make the Hack compiler.
|
|
#
?
Mar 25, 2017 15:46
|
|
|
Mniot posted:When I interviewed there last year, they said the Boston office had a lot of Facebook's version of SRE, security people, and some of the group that make the Hack compiler. Ah. OK. My info is not particularly up to date or detailed. Let's go with ^ then.
|
|
#
?
Mar 25, 2017 15:47
|
|
|
There are two distinct definitions about the term devops that makes any conversation about it meaningless unless you pick exactly what you're talking about. 1. Role-based definition stretching anywhere between "build bitch" to "sysadmin that doesn't run and scream from a command line" (yes, a lot of sysadmins never touched a script in their lives - I literally put them out of jobs) to "site reliability engineer." All of them imply someone that knows something more about code than a for loop and isn't afraid of getting their hands dirty with the realities of shipping serious software once you hit git push 2. An organizational / cultural style closely related to lean and Agile methodologies that breaks down organizational silos between roles in favor of alignment around services instead of function while typically blurring the responsibilities of engineers historically between a development and operations / infrastructure organization. Definition 1 preceded definition 2 because definition 2 is almost exclusively used by people trying to land consulting gigs for large, dysfunctional enterprises with technologies, persons, and methodologies stuck in the 90s back to the 60s with accompanying culture. These places after writing off these smaller companies for decades with a smug attitude are now looking for answers from all these startups and unicorns that grow much faster than them, deploy literally several orders of magnitudes of features more (and growing even further apart partly because of technical and cultural debt), and otherwise wipe the floor in stock and aggregate employee performance for the same business verticals. All of the definitions, however, mean "nothing like how people developed and ran software in the 90s" which is not really applicable to anyone that's worked in technology somewhere other than a standard low-performing shop with expected low morale and low "give a poo poo." Paolomania posted:So is "devops" the new "ops"? necrobobsledder fucked around with this message at 19:13 on Mar 25, 2017 |
|
#
?
Mar 25, 2017 19:09
|
|
|
Mniot posted:When I interviewed there last year, they said the Boston office had a lot of Facebook's version of SRE, security people, and some of the group that make the Hack compiler. Weird, I had a recruiter contact me about a SRE position that was in NYC, Seattle, or Menlo Park, but specifically not Boston even though I asked.
|
|
#
?
Mar 25, 2017 19:38
|
|
|
If SRE positions are essentially software engineer plus a pager then I would expect that there would be SREs at every office that has technical people. I mean, software engineers are going to be everywhere, they're not assigned to places based on job type but based product developed at each location, right? Why wouldn't SREs work the same way?
|
|
#
?
Mar 25, 2017 19:51
|
|
|
Blinkz0rz posted:Weird, I had a recruiter contact me about a SRE position that was in NYC, Seattle, or Menlo Park, but specifically not Boston even though I asked. Well this was from a FB recruiter, and they never said "SRE". They were telling me that the Boston office did things like figure out how they were going to build the 200th layer of caching around the core MySQL DB, or improve the deploy process that did all these automated roll-outs, canaries, and roll-backs, or come up with a "best practice" and then change all the lovely code to do it that good way.
|
|
#
?
Mar 25, 2017 20:26
|
|
|
So how many people here really know how a computer works, how a program you write results in magic happening in a computer's processor? When I graduated with a degree in CS, I thought I was smart. I got the best grades in my classes! I could kind of muddle through a dynamic programming algorithm or write a crappy ray tracer, but I didn't know anything about HTTP (what the hell is this "GET"? what's "POST"?). Later, I studied for interviews at big companies and realized I didn't really know anything about data structures or algorithms, either. I watched Tim Roughgarden's Coursera courses and for each topic, also learned from and did some practice problems from one of like 4 algorithm textbooks (they all had easily-comprehensible sections and incomprehensible sections, and none had easily-comprehensible sections for all topics). Then I thought I was good, but hey, someone said I should learn what an operating system does. I got a B somehow in my OS class, but didn't know anything about them. Found a book on them and read it about 3 years after taking the class. Similarly, didn't have any idea how TCP or IP worked. And now I've just realized, that while I have a vague notion of how an OS works, an OS is just one piece in the chain from some high-level language code (compiled) to assembly code (sort of interpreted by, sort of passed on by) to the OS which consists of more assembly, which is taken by the CPU and converted into actions between a bunch of registers and things, which are themselves implemented by "gates" or whatever that do magic, essentially. I had a digital logic class, an assembly language class, a computer architecture class, an operating system class, and a computer networking class, and I got As all but operating systems, where I got a B, but I never noticed how they were related and graduated without any understanding of how the code I write makes stuff do stuff. So I feel like I know basically nothing, but then I have a pretty good job at a major software company, and I'm like... does that mean most people working as "software engineers" know even less or have I just done a good job of faking it? Maybe I shouldn't have gotten hired but I passed because everyone wants to ask graph and tree traversal questions and nobody asks systems questions other than basic concurrency stuff you'd learn about in an operating systems class?
|
|
#
?
Mar 25, 2017 20:32
|
|
|
Very few people understand that stuff. Most of us can get by without knowing as long as we're willing to accept that some things just work in certain ways. Probably a lot of us these days are working in environments that are abstracted away enough from the physical machine that we don't even really see those parts. I went to school before most of what was in the previous paragraph was true, and as a result I did get a fair amount of training in what goes on under the hood. But nowadays it's mostly just interesting trivia. If you want a little bit more insight into how it all fits together, check out Computer Science from the Bottom Up. It won't tell you everything, but it will tell you lots of things, and then you'll have words to Google for the things you want to learn more about. Edit: It's also worth noting that nobody expects you to know too much coming out of an undergraduate computer science program. That's just to get you to the point where you have some basic knowledge and you've written some code. It's after you graduate that you start learning how to build software. ultrafilter fucked around with this message at 20:44 on Mar 25, 2017 |
|
#
?
Mar 25, 2017 20:42
|
|
|
Yeah I still feel a distinct lack of operating systems and low-level knowledge because everything I work with is a god drat framework (Spark now, which is one of the worst tbh) and while I've obviously learned a ton in my jobs, a comprehensive operating systems class (or systems programming in a Unix environment, etc) is something I wish I could find on like Coursera. Come to think of it - there has to be somewhere that to rate these types of courses right? Say I want to do a comprehensive functional programming course, there should be a place I can go to find the best freely available resource for that or something. (I say courses because working 45 hours a week, I need the structure of a course to motivate me) Good Will Hrunting fucked around with this message at 20:57 on Mar 25, 2017 |
|
#
?
Mar 25, 2017 20:53
|
|
|
Mniot posted:Well this was from a FB recruiter, and they never said "SRE". They were telling me that the Boston office did things like figure out how they were going to build the 200th layer of caching around the core MySQL DB, or improve the deploy process that did all these automated roll-outs, canaries, and roll-backs, or come up with a "best practice" and then change all the lovely code to do it that good way. Facebook doesn't have SREs, the equivalent is probably PEs (Production Engineer)?
|
|
#
?
Mar 25, 2017 21:22
|
|
|
Good Will Hrunting posted:Come to think of it - there has to be somewhere that to rate these types of courses right? Say I want to do a comprehensive functional programming course, there should be a place I can go to find the best freely available resource for that or something. (I say courses because working 45 hours a week, I need the structure of a course to motivate me) Class Central is the closest thing I'm aware of.
|
|
#
?
Mar 25, 2017 22:22
|
|
|
Doctor w-rw-rw- posted:Facebook doesn't have SREs, the equivalent is probably PEs (Production Engineer)? Yeah that was the department/org/group/whatev but it's basically the same thing so I didn't want to split hairs.
|
|
#
?
Mar 26, 2017 03:36
|
|
|
This week will be two months at my new job. I've seen virtually no improvement from management in terms of defining the direction of our project, goals, resource allocation or any sort of medium-term roadmap. We routinely spend 45 minutes to over an hour in our daily stand-ups arguing over trivial bullshit. I have no idea what one of the senior engineers does to be honest. The flow of any sort of product or business analyst to dev is non-existent. There's basically nobody speccing these stories and the person most involved with that (my boss, the VP of Engineering) is only very tangentially involved in this project because he's busy managing 10+ people, other projects, and working with non-engineering management. He also spends like, a ton of time coding and in a terminal? Timezones are also an issue. There are two other managers who clearly have no idea what the business requirements are and one attempts to come up with answers to questions we have that are frequently guesses, it seems. Another problem though is that a lot of the things my manager wants to accomplish strike me as ridiculously over-engineered. He seems to want to engineer our own version of Kafka to use with Spark Streaming because Google Cloud doesn't have a managed Kafka service. I'm still struggling to understand why we need to do this as opposed to using another managed service they offer which could solve our problem (which they can - I researched this myself during downtime) and his answer to my proposition was "It's not Kafka and that's why we aren't using it". is this bad luck or am I just learning the way of software development at start-ups?
|
|
#
?
Mar 27, 2017 15:28
|
|
|
The Something Awful Forums > Discussion > Serious Hardware / Software Crap > The Cavern of COBOL > Oldie Programming: He seems to want to engineer our own version of Kafka
|
|
#
?
Mar 27, 2017 15:42
|
|
|
|
| # ? Apr 19, 2024 07:13 |
|
|
Good Will Hrunting posted:This week will be two months at my new job. I've seen virtually no improvement from management in terms of defining the direction of our project, goals, resource allocation or any sort of medium-term roadmap. We routinely spend 45 minutes to over an hour in our daily stand-ups arguing over trivial bullshit. I have no idea what one of the senior engineers does to be honest. Failure to properly plan or delegate is pretty standard in startups.
|
|
#
?
Mar 27, 2017 16:02
|
|
