Register a SA Forums Account here!
JOINING THE SA FORUMS WILL REMOVE THIS BIG AD, THE ANNOYING UNDERLINED ADS, AND STUPID INTERSTITIAL ADS!!!

You can: log in, read the tech support FAQ, or request your lost password. This dumb message (and those ads) will appear on every screen until you register! Get rid of this crap by registering your own SA Forums Account and joining roughly 150,000 Goons, for the one-time price of $9.95! We charge money because it costs us money per month for bills, and since we don't believe in showing ads to our users, we try to make the money back through forum registrations.
 
  • Post
  • Reply
GreenNight
Feb 19, 2006
Turning the light on the darkest places, you and I know we got to face this now. We got to face this now.

CommieGIR posted:

I too can do useless changes 200 times a day.

No no no. You tell someone else to do useless changes. My forte.

Adbot
ADBOT LOVES YOU

Cimber
Feb 3, 2014
I have to wonder, all these companies that brag that they release to prod 200+ times a day, what the gently caress are they releasing so much, and how the hell does CI/CD ensure that the code changes are not causing minor fuckups that might not be noticible for a few hours/days/weeks until suddenly a major outage happens and they have to roll back to code a month old?

Thanks Ants
May 21, 2004

#essereFerrari


It will be things like a team correcting a typo in a UI element somewhere getting deployed automatically after passing the various checks. It’s not going to be people launching hundreds of features a day.

tokin opposition
Apr 8, 2021

I don't jailbreak the androids, I set them free.

WATCH MARS EXPRESS (2023)
Gonna become an AI Engineer SRE CTO where we run dozens of chatgpt instances all making code changes at random and we *checks bizweek* "red team" them against one another until the codebase continuously improves itself

I will, of course, be laying off the entire engineering org and taking that budget as my salary

Twerk from Home
Jan 17, 2009

This avatar brought to you by the 'save our dead gay forums' foundation.

Cimber posted:

I have to wonder, all these companies that brag that they release to prod 200+ times a day, what the gently caress are they releasing so much, and how the hell does CI/CD ensure that the code changes are not causing minor fuckups that might not be noticible for a few hours/days/weeks until suddenly a major outage happens and they have to roll back to code a month old?

Feature flags, you don't roll it back you just turn the problematic feature off until you can fix it.

Cimber
Feb 3, 2014

Twerk from Home posted:

Feature flags, you don't roll it back you just turn the problematic feature off until you can fix it.

Uhh, yeah, and how exactly does that work in finance? "Oh, sorry Federal Reserve, we can't report our reserve requirement compliance because well, there's a bug in our software, you see? We had to turn off that feature until the next sprint cycle when we can budget story points."

Internet Explorer
Jun 1, 2005





This conversation is totally cool here but I just wanted to advertise the CI/CD thread -
https://forums.somethingawful.com/showthread.php?threadid=3695559

Cimber
Feb 3, 2014

Internet Explorer posted:

This conversation is totally cool here but I just wanted to advertise the CI/CD thread -
https://forums.somethingawful.com/showthread.php?threadid=3695559

That thread does not have enough posts/day to be considered accurate.

Twerk from Home
Jan 17, 2009

This avatar brought to you by the 'save our dead gay forums' foundation.

Cimber posted:

Uhh, yeah, and how exactly does that work in finance? "Oh, sorry Federal Reserve, we can't report our reserve requirement compliance because well, there's a bug in our software, you see? We had to turn off that feature until the next sprint cycle when we can budget story points."

If it's something absolutely critical do or die then yeah your tests had better make sure that what you are deploying does the thing.

I get that bragging about deploy frequency sounds silly, but most of us have worked in places where deployments are rare and awful and I would absolutely never go back to that.

If you are deploying rarely and it's a big event, you can never get good at it.

CLAM DOWN
Feb 13, 2007




Internet Explorer posted:

This conversation is totally cool here but I just wanted to advertise the CI/CD thread -
https://forums.somethingawful.com/showthread.php?threadid=3695559

I used to check that thread but no one else seems to use an Azure PaaS/AzDO stack, also methanar used to frequent that thread lmao

Internet Explorer
Jun 1, 2005





well look I'm a mod not a miracle worker

jaegerx
Sep 10, 2012

Maybe this post will get me on your ignore list!


Internet Explorer posted:

well look I'm a mod not a miracle worker

uh, you work in IT.

Everything we do is a miracle

Cimber
Feb 3, 2014

Internet Explorer posted:

well look I'm a mod not a miracle worker

Do i need to raise the severity of my ticket, buster?

GreenNight
Feb 19, 2006
Turning the light on the darkest places, you and I know we got to face this now. We got to face this now.

Escalate - work stoppage

CommieGIR
Aug 22, 2006

The blue glow is a feature, not a bug


Pillbug

Cimber posted:

Do i need to raise the severity of my ticket, buster?

Its still within its SLA *taps service now*

Cimber
Feb 3, 2014

CommieGIR posted:

Its still within its SLA *taps service now*

I can raise this to a Major Incident and I'm not afraid to do it!

Organic Lube User
Apr 15, 2005

Why is everything related to SAP so awful and dry? It's been around, largely unchanged, for what, 30 years now? Why is all the documentation at every company I've ever been at seem like it was a powerpoint Google translated from the original German, and is also missing an accompanying presentation for actual context?

I've got like two more days of training and tests for this system that I'm told I will access maybe once a year.

Submarine Sandpaper
May 27, 2007


The security side of SAP is awful. You have to remember a bunch of tcodes that mean nothing.

Su is user, pa is HR, SE is tables.

I pitty anyone who goes into it

vanity slug
Jul 20, 2010

Software Against People

Hughmoris
Apr 21, 2007
Let's go to the abyss!

Submarine Sandpaper posted:

The security side of SAP is awful. You have to remember a bunch of tcodes that mean nothing.

Su is user, pa is HR, SE is tables.

I pitty anyone who goes into it

Su(do) Users
your pa(L) in HR
s(pecial) e(xcel) tables


Easy peasy

Nuclearmonkee
Jun 10, 2009


Cimber posted:

I have to wonder, all these companies that brag that they release to prod 200+ times a day, what the gently caress are they releasing so much, and how the hell does CI/CD ensure that the code changes are not causing minor fuckups that might not be noticible for a few hours/days/weeks until suddenly a major outage happens and they have to roll back to code a month old?

If your automated tests are good enough, then most of those don’t make it to prod. There’s always a hole somewhere though and there’s some risk in making changes no matter how you do it.

The more intricate tests are usually put there because of some event or another.

And 200 changes a day is just some manager looking at how many commits they average per day in their repo, not the content of those commits. Minor code fixes, “added comment”, and other crap will account for most of those. No one makes that many major application changes in one day just little things with some microservice or another.

Nuclearmonkee fucked around with this message at 14:20 on Apr 1, 2024

rafikki
Mar 8, 2008

I see what you did there. (It's pretty easy, since ducks have a field of vision spanning 340 degrees.)

~SMcD


Pretty sure this isn’t an April fool’s joke https://www.reuters.com/technology/microsoft-separate-teams-office-globally-amid-antitrust-scrutiny-2024-04-01/

Nuclearmonkee
Jun 10, 2009



This will somehow involve a net increase to their pricing structure, and they will just blame the EU for it.

CommieGIR
Aug 22, 2006

The blue glow is a feature, not a bug


Pillbug
SAP is a good example of why you don't like MBAs determine how a coding platform will work.

Thanks Ants
May 21, 2004

#essereFerrari



Can't really argue with that, all our customers who use Teams have "we already have the licenses for it" as their reason.

LochNessMonster
Feb 3, 2005

I need about three fitty


Cimber posted:

Uhh, yeah, and how exactly does that work in finance? "Oh, sorry Federal Reserve, we can't report our reserve requirement compliance because well, there's a bug in our software, you see? We had to turn off that feature until the next sprint cycle when we can budget story points."

There’s this concept called “testing” which is fairly important in CI. Doing that properly helps if you want to do the whole CD part.

Systems being important does not make CI/CD a more or less viable concept. In all fairness, I’d rather work on a system that’s using CI/CD to deploy then one that requires manually steps. Especially in finance where a fully automated audit trail is so much more convenient then you trying to come up with which steps you did exactly during that 11pm to 6 am release. And where this deviated from that 189 page runbook with 12 addendums which still was outdated/incomplete after it being prepared for by 7 people over 3 months.

Not trying to be an rear end in a top hat here, but if you’re (this) vocal on sticking to how you’ve always worked as opposed to trying to improve your way of working, it might come off as old-man-yelling-at-the-cloud.jpg to your interviewers. I’m not saying that played a role as I obviously wasn’t interviewing you, but I’ve been through several reorgs and people who are less open to change are usually the ones that are let go.

Vampire Panties
Apr 18, 2001
nposter
Nap Ghost

Thanks Ants posted:

Can't really argue with that, all our customers who use Teams have "we already have the licenses for it" as their reason.

Its true, but the new licenses for standalone teams are cheap as chips, and anyone on an enterprise agreement can keep renewing it. Since 90% of orgs see collaboration as a cost center, and migration away from teams is effectively throwing all of your chat/phone history in the fire and starting over, this won't really change much. Its a chink in MS's armor for orgs that somehow haven't made the jump to the cloud yet, but the pricing is still so competitive that I'm not sure that it matters. Also I'm not clear if Office gets appreciably less expensive w/o teams that orgs would consider multi-vendor solutions.

Maybe a big win for Zoom, since they have word processor and email clients built into their platform. The extra 5bux for MS Teams would push Zoom into being price competitive when you're all in, since the now-13bux for MS Teams Phone doesnt include any PSTN connectivity, but MS also has the shared calling package so multiple users can split a single line.

I'm really curious what this will do to the Fed market. Everyone with a .gov email address has been told for years that all of their MS Teams is free, included in department-wide enterprise agreements, and a lot of places are migrating to MS teams phone specifically because of the reduced cost of licensing. Seems like a lot of multi-year migration plans will be put on hold while they figure out what the long term costs will be, especially with the aforementioned teams-dumpster-fire migration policies. Huge loving :lol::lol::lol: if a buncha ancient telco dudes get cold feet and decide to stick with Cisco (specifically Webex dedicated instance) and just keep doing what they're doing forever

Cimber
Feb 3, 2014

LochNessMonster posted:

There’s this concept called “testing” which is fairly important in CI. Doing that properly helps if you want to do the whole CD part.

Systems being important does not make CI/CD a more or less viable concept. In all fairness, I’d rather work on a system that’s using CI/CD to deploy then one that requires manually steps. Especially in finance where a fully automated audit trail is so much more convenient then you trying to come up with which steps you did exactly during that 11pm to 6 am release. And where this deviated from that 189 page runbook with 12 addendums which still was outdated/incomplete after it being prepared for by 7 people over 3 months.

Not trying to be an rear end in a top hat here, but if you’re (this) vocal on sticking to how you’ve always worked as opposed to trying to improve your way of working, it might come off as old-man-yelling-at-the-cloud.jpg to your interviewers. I’m not saying that played a role as I obviously wasn’t interviewing you, but I’ve been through several reorgs and people who are less open to change are usually the ones that are let go.

Oh, I'm not at all disagreeing that CI/CD have huge advantages over say, quarterly releases, it's going back to my earlier question of how do we know that the minor change isn't causing a problem that might only appear once a month or some other non obvious timeframe? If a developer uses a float vs a double in a variable for example and interest rates that might need to go out to X number of decimal places might get lopped off at X-3, causing a mismatch between what we report and someone else reports, but that mismatch only appears after 6 months?

[edit] Buuuut, I'm a unix SA. I've never written huge code like this. I've never been expected to write huge code like this and while perhaps there are devopment mechanisms that would identify the problem I'm addressing, I'm suddenly being held to a standard that has never been held to me before, to 'be a software developer because Google thought it was cool'.

Cimber fucked around with this message at 15:23 on Apr 1, 2024

LochNessMonster
Feb 3, 2005

I need about three fitty


Cimber posted:

Oh, I'm not at all disagreeing that CI/CD have huge advantages over say, quarterly releases, it's going back to my earlier question of how do we know that the minor change isn't causing a problem that might only appear once a month or some other non obvious timeframe? If a developer uses a float vs a double in a variable for example and interest rates that might need to go out to X number of decimal places might get lopped off at X-3, causing a mismatch between what we report and someone else reports, but that mismatch only appears after 6 months?

[edit] Buuuut, I'm a unix SA. I've never written huge code like this. I've never been expected to write huge code like this and while perhaps there are devopment mechanisms that would identify the problem I'm addressing, I'm suddenly being held to a standard that has never been held to me before, to 'be a software developer because Google thought it was cool'.

This really is where testing plays a big role. If you want to properly/safely deploy to production you want to have all sorts of automates tests (unit, integration, regression, load, etc).

It’s unreasonable for your employer to expect you to know this if you’ve never been exposed to these concepts before. It’s like asking a a car mechanic to do an aviation mechanic interview.

Cimber
Feb 3, 2014
I can't disagree that its an unreasonable expectation.

Also in reality, how often do SREs actually do deployments like this? Do they actually write code, or are most of the time are they busy with other things like monitoring/dashboards/RCAs like I'm already doing. :P

[edit] And for my curiosity, how do they do all these tests so fast if the expectation is to deploy so frequently?

Cyks
Mar 17, 2008

The trenches of IT can scar a muppet for life

Nuclearmonkee posted:

This will somehow involve a net increase to their pricing structure, and they will just blame the EU for it.

Yep. It’s $2-$3 more a month per licensed user now.

If you already have an active subscription you are good. Any new customers or new subs (for example, switching from partner to direct is a new sub) will be at the newer price. New pricing also only impacts Enterprise licenses (e1, e3, e5 for both office and m365).

If you don’t use teams it’s actually cheaper, but who isn’t using teams with an e3/e5 license?

Cyks fucked around with this message at 15:52 on Apr 1, 2024

Darchangel
Feb 12, 2009

Tell him about the blower!



LO-loving-L - we changed to Teams because "$free" and Zoom kept jacking up their price during COVID.


Stellantis pulled an incredibly assholish move to lay off tech workers (because obviously Stellantis is a car company and has no need of tech and softweare engineers...) - they ordered the workers to WFH for a day and then fired them remotely.
https://www.msn.com/en-us/money/com...all/ar-BB1kK9mv

Welcome to late-stage Capitalism?

LochNessMonster
Feb 3, 2005

I need about three fitty


Cimber posted:

I can't disagree that its an unreasonable expectation.

Also in reality, how often do SREs actually do deployments like this? Do they actually write code, or are most of the time are they busy with other things like monitoring/dashboards/RCAs like I'm already doing. :P

[edit] And for my curiosity, how do they do all these tests so fast if the expectation is to deploy so frequently?

CI and CD are 2 different entities. So the testing part is seperate from the deployment part.

Tests are usually run only for the parts that have changed. They are also run locally and in yourfeature branch so you’re already aware things are broken and you can’t release your garbage code to prod. The idea is that the sooner you realise you made a mistake, the easier it is to fix (short feedback loops).

If you would like to read about the theory I can recommend the books from Gene Kim on the concepts. The DevOps Handbook, The Phoenix Project and The Unicorn Project. You’ll probably recognize a lot of probelms you’ve encountered yourself as well.

Cimber
Feb 3, 2014
I've also heard that there are two different CDs, continuous delivery and continuous deployment which are two distinct things. So theoretically testing might include scenarios where it's calculating the results as if it was 90+ days out to see if the interest rate reports are valid?

I guess it gets into the known unknowns and unknown unknowns issue which seems to be a rabbit hole way beyond this thread.

tokin opposition
Apr 8, 2021

I don't jailbreak the androids, I set them free.

WATCH MARS EXPRESS (2023)
My coworker is currently screaming and throwing things in her office loud enough to hear it thru the wall and a closed door about a phone ticket with an external vendor, hasn't talked to me yet but I can already tell today is going to be Great.

Haha I'm in danger

GreenNight
Feb 19, 2006
Turning the light on the darkest places, you and I know we got to face this now. We got to face this now.

tokin opposition posted:

My coworker is currently screaming and throwing things in her office loud enough to hear it thru the wall and a closed door about a phone ticket with an external vendor, hasn't talked to me yet but I can already tell today is going to be Great.

Haha I'm in danger

Who the gently caress goes to work and decides that today is the day I'm gonna have a temper tantrum with a vendor?

Zorak of Michigan
Jun 10, 2006


GreenNight posted:

Who the gently caress goes to work and decides that today is the day I'm gonna have a temper tantrum with a vendor?

I'm withholding judgement until I hear who the vendor is and whether the tantrum got results.

Thanks Ants
May 21, 2004

#essereFerrari


It's a phone ticket so it's probably justified. You should get some sort of presidential medal if you manage to deal with an incumbent carrier without showing up at their offices to exert violence.

Submarine Sandpaper
May 27, 2007


For me it's what days I don't

Adbot
ADBOT LOVES YOU

Vargatron
Apr 19, 2008

MRAZZLE DAZZLE


tokin opposition posted:

My coworker is currently screaming and throwing things in her office loud enough to hear it thru the wall and a closed door about a phone ticket with an external vendor, hasn't talked to me yet but I can already tell today is going to be Great.

Haha I'm in danger

Oh I've been there before. Spiked a company phone in a parking lot once after a heated argument with my boss.

drat I'm glad I don't work in automotive anymore.

  • 1
  • 2
  • 3
  • 4
  • 5
  • Post
  • Reply