|
I work in a legacy industry and my Docker exposure is crap, so... is there a joke buried in there, or is there some good reason why you need to build and deploy containers within Docker instead of without?
|
# ¿ Mar 28, 2018 19:44 |
|
|
# ¿ May 15, 2024 02:54 |
|
It's nice to hear other people saying that stuff. Every time we push back at someone's demand for a 16 vcpu 128gb RAM VM around here, I get a lot of whining that boils down to, "but Mom would give us a VM that size!" Then we put them on a 2x16, evaluate usage, add 8 more gigs of RAM, and find that it runs fine and the requirements were entirely fictional anyway.
|
# ¿ May 4, 2018 20:43 |
|
My first Linux was Centos 7.3. ... but my first UNIX was SunOS 4.1.3.
|
# ¿ May 30, 2018 15:24 |
|
This is a tangled web because harsh penalties or huge bonds push business toward a few large firms that can afford them. That in turn creates a situation where you only have a couple megacontractors. Those large contractors can afford to eat small penalties, and nobody has the will to impose large ones, because you might put one of your 2-3 megacontractors out of business, and then you've just made the oligopoly worse. Off the top of my head, the solution that suggests itself is to add a third party responsible for auditing progress and ensuring the success of the contract. They'd need to be well capitalized and to understand the subject matter, but they wouldn't be involved in actual service delivery. They'd rate the proposals and the bidders themselves and provide quotes for insuring the success of each bid. Those insurance costs could then be added to the actual bid costs to come up with a final price. Large companies with bad practices would still be in the running so long as they could afford insurance rates, but smaller and/or newer players could compete fairly for business provided they could demonstrate competence to the insurers. I have no idea how you could get that off the ground, though.
|
# ¿ Jun 12, 2018 19:16 |
|
I believe Chairman Kaga has already ruled in favor of bell peppers, case closed.
|
# ¿ Jul 25, 2018 15:06 |
|
Of all the reasons I evangelize containers around the office, ease of backing out a new release is #1.
|
# ¿ Aug 31, 2018 14:56 |
|
Agrikk posted:Their herp-derp incompetence makes me die a little bit more in anguished outrage every time we field a critical case that is 100% their ignorance and 0% our platform. One of my career life goals is to be in a situation where I can respond to one of those tickets with "wow, it really sounds like you hosed up. All our processes operated within our SLAs and so there is nothing we can do for you. Would you like to meet with someone to discuss best practices and how you could stop your future dumb decisions from turning into outages?"
|
# ¿ Sep 13, 2018 20:00 |
|
It's not clear to me that Renegret meant "goddamned CEO" literally, and if not, that puts a rather different construction on things. Depending on the size of the company, it doesn't always make sense for senior leadership to have the same process as everyone else. If I was running a multinational, I'd set up a frickin' IT concierge service for the C-level executives. Partially that's because their time is insanely valuable and I wouldn't want them waiting on hold, and partially because anyone touching their IT assets needs to know that their daily activities create documents with special legal status. Having them run into some IT intern who decides to wipe and reimage their laptop could be... bad. Having a C-level decide that IT isn't responsive enough and they need their own shadow IT would be infinitely worse.
|
# ¿ Dec 6, 2018 21:56 |
|
Back when I was at Umich, they not only had lavish vacation, but a very simple rule for accumulation: if you had more than 18 months' vacation accumulated (and I think we were at 2 days PTO per month, so that's 36 days of PTO) you stopped accumulating it. So there was no arbitrary "everyone use up their PTO by 12/31" nonsense, but individuals who didn't take PTO eventually lost it. That seemed like a sufficiently fair protection for the University's financial concerns. When I was a manager at my current gig, which has use it or lose it PTO that includes sick time, everyone naturally went in to December with at least a week of PTO left. After all, you don't want to risk getting sick at the end of the year and having to take unpaid leave. I ended up instituting a tiered vacation system for my team. Tier 1 meant you were well and truly gone. Tier 2 meant you had no responsibilities except to stay in pager range and be able to return to work in 2 hours if I paged you. That got me the leeway to put up to 1/3 of my people on tier 1 vacation and another 1/3 on tier 2, with only the remaining third in the office. My management didn't love having 2/3 of the staff out, but since I could go from 1/3 working to 2/3 working in 2 hours if we had a major incident, they allowed it. Which is still just a good workaround for a dumb policy.
|
# ¿ Dec 11, 2018 23:02 |
|
Agrikk posted:Imagine being “the DNS guy”. I used to be that guy, so for me, it's more like "imaging ignoring 15 years of advances in enterprise computing and thinking it gave your job security." That's one of those plans that sounds great until it stops working suddenly, completely, and forever.
|
# ¿ Dec 13, 2018 18:05 |
|
AlexDeGruven posted:Recruiter ping this morning. Sounded interesting. Asked for more info. They forwarded a job description. Way more windows-ey than I'm comfortable with, but nbd. Then I see it: I'm only at 90% nope, with the other 10% being "well, at least they know that it's happening and they're honest about it."
|
# ¿ Dec 13, 2018 21:02 |
|
Proteus Jones posted:10 second walk into home office after breakfast, shower and taking care of pets. And that's the winter wardrobe. Summertime, if I happen to be certain my kid won't be around, it's sometimes just underwear.
|
# ¿ Dec 29, 2018 03:40 |
|
Bunch of prudes in this thread.
|
# ¿ Dec 29, 2018 07:42 |
|
incoherent posted:Yolo'in update 4 of veeam rn to get at scale out storage w/ AWS. Wish me luck. I would love a trip report once you have AWS added to your SOBR and data flowing to it.
|
# ¿ Jan 24, 2019 00:51 |
|
Glocks don't have safeties, so you can stop pondering that part of the question.
|
# ¿ May 1, 2019 15:56 |
|
A ticket came in: we started running a database workload that does nothing but full table scans all through the day and all through the night, and now the database is slow. Can you put us on different hardware that will make it fast? No; or rather, I can, but you can't afford it. Perhaps you might, you know, not do that so much, using indexes or some other crazy database magic.
|
# ¿ May 8, 2019 22:29 |
|
nexxai posted:Is this person/team at least using a read replica for whatever they're doing, so that they don't slow it down for everyone else? I'm having lamentable difficulty even finding out exactly what they're doing. Thankfully it's in our non-production environment, so it's not impacting production. There are multiple instances of it, too, and I'm concerned that in a few places it may actually be swamping the aged FC infrastructure for those machines. If it was in production, at least we'd have tools that I could use to analyze the queries and tell them to stop their nonsense in a more authoritative way.
|
# ¿ May 9, 2019 04:24 |
|
I don't want to imagine someone who'd take a work asset with them on an international vacation.
|
# ¿ Jun 6, 2019 15:10 |
|
Wibla posted:I've done it a few times, was nbd. I was thinking in my USA context, where international travel always means border checkpoints, the risk of intrusive searches, and in general, feels like a big deal. If my vacation is a big deal, I'm not bringing work with me. Actually, I can't recall ever bringing my work laptop with me on a non-work trip, but it's easier to imagine doing so if the trip is just a couple hours from home and there's minimal hassle involved.
|
# ¿ Jun 6, 2019 16:56 |
|
Dirt Road Junglist posted:Third dumbest? Now I wanna hear numbers one and two. I did suggest reimaging, but a VP told me, "absolutely not an option." Technically, we don't have the resources what with most of the competent support folks manning the Cloud Prom in SF this week, so he's not entirely wrong. I just wish he'd stop telling me no every time I suggest reimaging for poo poo that obviously would be easier to deal with by reimaging. Do you have the sort of relationship where you could ask him, or get someone in your management chain to ask him, what they could do to increase his confidence in the imaging process?
|
# ¿ Nov 22, 2019 21:24 |
|
I've been full-time WFH since 2003, and I love it, but I wouldn't want to start a brand new job in a brand new firm without there being an office I could go into. I think my current gig has worked as well as it has because I've always worked at least 50% of the time with people I've had established relationships with. In a brand new situation, I'd want at least a couple weeks of face-to-face interaction to build on before I started doing everything remote. Maybe in a better company, I could rely on written procedures and practices rather than relationships, but probably not.
|
# ¿ Dec 1, 2019 21:28 |
|
AlexDeGruven posted:I mean, I'd live over there on SV wages I do live on the east side but I wouldn't choose Roseville.
|
# ¿ Feb 3, 2020 20:04 |
|
AlexDeGruven posted:Didn't/don't you work in Ann Arbor? That's right. East side compared to GR but not Detroit or anything.
|
# ¿ Feb 3, 2020 21:49 |
|
Sirotan posted:Sweet. How do you want to coordinate this? Email chain? Goon meets thread (lol)? I'm also on Telegram if anyone uses that. Despite having recently posted about how I love working from home because I don't have to interact with humans, hell yeah I'm in.
|
# ¿ Feb 5, 2020 01:48 |
|
u brexit ukip it posted:I would prefer this over my situation where it has taken IT about three months to raise 5 VMs in Azure. This is why I laugh at people in my org who think that moving to the cloud would make us more nimble. Sure, our process for getting new VMs from our on-prem setup could be better. Sure, when you go to the cloud, it's instant gratification. For us, that would last until the first bill came due. Then it would be "any increase in cloud resources requires approval from the director of cloud cost control" and a whole new sucky request process. If you don't have employees you can trust to be sensible, and a willingness to delegate authority to them, then changing infrastructure cannot save you.
|
# ¿ Feb 6, 2020 20:53 |
|
Wild. I always prefer to put team meetings on Wed, since it's the day people take off the least.
|
# ¿ Mar 16, 2020 22:01 |
|
larchesdanrew posted:I close on my house on July 8th. My last day in this place is June 30th. I submitted my notice and they began advertising for my position. Hell yeah!
|
# ¿ Jun 10, 2020 01:28 |
|
Break a leg, Larches. Take what you can. Give nothing back.
|
# ¿ Jun 22, 2020 20:18 |
|
larchesdanrew posted:A job came in Outstanding.
|
# ¿ Jun 24, 2020 17:59 |
|
devmd01 posted:That’s my favorite interview question: “tell me about a time you caused an outage, the impact to the business, the recovery process, and what you learned from it?” Had to think for a minute but I remember the one time I had been using the Solaris modular debugger on the kernel of a production mission-critical DB server to get performance data so I could troubleshoot a problem. I left the window open, thinking I'd close it once I was sure the problem was solved. At the end of the day, when I wanted to shut down, I went and typed "exit" into every open window, disconnected my screen session, and logged out. The next morning I re-opened screen and saw that my mdb session was still open. Oh, right, exit isn't how you exit mdb on Solaris. Control-d is how you do that. I pressed control-d. What I found out later is that exit means nothing to mdb in and of itself, but it was in the command buffer when I hit control-d, so the mdb just passed that along to the kernel. Which is how I killed the kernel of a production mission-critical Oracle database first thing in the morning. From then on, I only ever touched mdb by using echo "exactly what I want, and yes, I checked twice" | mdb -k.
|
# ¿ Jul 9, 2020 20:24 |
|
Did he know the gun holder personally? The mindset of "I know him and like him, so what he's doing is cool" is common, hard to break, and it makes someone useless as a safety officer. Of course, if people weren't useless at their jobs, we wouldn't have this thread.
|
# ¿ Aug 6, 2020 19:19 |
|
I love automation. When I was setting up our first Puppet POC, I worked a couple weekends on it for free, because I was into it and couldn't let it go. Other exotic problems, maybe interesting. Your server is slow because your SQL is badly written and had a cost of about a million on a four core guest? I can barely get the bat off my shoulder.
|
# ¿ Aug 13, 2020 01:28 |
|
Oh, to be that young. In teaching Puppet stuff to other members of my org, I have shown them how to do things in VS Code, which has nice Puppet integrations, but I find that I'm just more fluent with vim, and also that things seem to go a lot faster without the VS Code layer between me, my files, and git.
|
# ¿ Aug 25, 2020 20:04 |
|
I have resigned myself to shitposting in four minute chunks.
|
# ¿ Nov 3, 2020 14:39 |
|
Depends what you compare them to.
|
# ¿ Dec 6, 2020 05:52 |
|
IPXes were useful for Xkernel for a little while, but by the late 90s they weren't even toys.
|
# ¿ Dec 8, 2020 03:06 |
|
Better yet, say what you need the attention for, so that the recipient can prioritize.
|
# ¿ Feb 10, 2021 14:58 |
|
One of my hats is infrastructure architect, with a particular focus on UNIX/Linux. If someone's doing a new project on our predominantly on-prem world that requires something more than a garden-variety VM deployment, and it's going to run Linux, they almost always come to me. There's another architect with a stronger Windows background who gets those requests. Except that he apparently decided to go around me for a Linux project because he was in a hurry, and now he's shocked, simply shocked, to discover that the standard Dell PCI RAID controller cannot create a RAID5 array out of NVMe drives. I have not yet found ways to ask "why did you think that was possible" and "why did you think that would be a good idea" that passed the rear end in a top hat test. I have to confess that I'm not trying real hard to pass said test because I do, in fact, think the guy is an rear end in a top hat.
|
# ¿ Mar 10, 2021 18:04 |
|
Look, if you don't make my application's VM a 32x128, how will I show that it's more important than my rival's app running on a 24x96 VM?
|
# ¿ Apr 6, 2021 01:25 |
|
|
# ¿ May 15, 2024 02:54 |
|
RAM-wise, I'll draw the line at anything using more than about an eighth of the host's total memory.
|
# ¿ Apr 7, 2021 04:00 |