|
I work in infrastructure. E-mails are coming in about the new KPTI bug. Looks like it's going to be worse than the old Pentium floating point bug. This will be a fun few weeks. 
|
#
¿
Jan 3, 2018 10:48
|
|
|
|
| # ¿ Apr 25, 2024 06:24 |
|
|
Sickening posted:So basically this shortcut has given intel a massive competitive advantage for the last decade in performance? It's not really a shortcut, it's just architecture that they didn't realise (at the time) could be leveraged this way. Essentially the issue exists because when one of the affected chips has to do a calculation that has to be handed off to the kernel, it keeps the kernel loaded into temporary memory to make the processing faster rather than having to reload the kernel for every subsequent calculation (I think; it's a bit over my head as well tbh). Some boffin somewhere figured out a way to exploit the kernel while it is loaded into temporary memory, allowing them to do all kinds of fuckery. The fix means that Intel chips will now have to call the kernel up every time it is needed rather than keep it in temporary memory which will slow down processing time hugely. Honestly, I wouldn't be too worried about regular desktop computers since they don't really see the number of IOPs where this will really be a problem. The big issue will be seen on clusters where IOPs are required at a premium and things like SQL processing. That is why AWS and Microsoft will be so badly impacted because their butt processing is massive. I'm sure they will be able to brute force spend their way out of it, however a 30% slowdown across their entire estate is the thing of nightmares. Edit: Also this only seems to be affecting CPU architectures in between Sandy Bridge and Coffee Lake (so like the 3000 series to the 7000 series?), so if you are still running a 2550k on your home PC like I am then you shouldn't see any problems. Dravs fucked around with this message at 16:15 on Jan 3, 2018 |
|
#
¿
Jan 3, 2018 16:12
|
|
|
Jaded Burnout posted:When you say "in between" do you mean inclusive? As 3000 series are Sandy Bridge. There was a tool that someone wrote to find out what is affected. I will try to find it again and edit this post with it. Edit: Bear in mind that nothing is gospel at the moment, there is so much speculation and chaos we might as well be running around in Hyboria being chased by giant dick eating lizards. Edit: Oh God, the exploits are starting to appear, AMD and Arm are also affected it seems? http://www.theregister.co.uk/2018/01/04/intel_amd_arm_cpu_vulnerability/ Also, it looks like the CEO of Intel dumped as much of his stock as possible just before Christmas. There is still an embargo on the actual information, but now I have seen some sources that all chips from the Pentium Pro onwards are affected, Jesus.  From that article: quote:Finally, if you are of the opinion that us media types are being hysterical about this design blunder, check this out: CERT recommends throwing away your CPU and buying an non-vulnerable one to truly fix the issue. lmao Pretty good article here if anyone wants to nerd out about it: https://arstechnica.com/gadgets/2018/01/whats-behind-the-intel-design-flaw-forcing-numerous-patches/ Update: Just got some info from a security tech at one of our vendors. Some choice quotes: quote:Summary quote:Scope
Dravs fucked around with this message at 11:44 on Jan 4, 2018 |
|
#
¿
Jan 4, 2018 09:42
|
|
|
Happiness Commando posted:What would cause a Server 2012 machine to insist that no logon servers are available to service a logon request, even though the firewall insists that traffic is being passed from that machine to the DC and back? Local accounts are disabled so there's literally no other way to get in short of maybe inserting a bootable ISO and rejiggering the local accounts. I dont know if the machine is in production, though, and can't necessarily bounce it. As has been said it has probably just lost the trust relationship. If you viewed the event log you would probably see errors about the RPC server being unavailable. This is just a Windows bug since forever and is the reason why you should always have a plan B to log onto a server (not disabling all of the local logon accounts would be a good way to start). How important is the server, can you just rebuild it?
|
|
#
¿
Feb 26, 2018 12:20
|
|
|
Avenging_Mikon posted:Oh, that's totally from forcing it in upside down. I think this is correct. Someone has tried to plug in a USB device and when finding it doesn't quite fit, instead of trying it the other way round, they have just rammed it harder and harder until it fit in the hole (that's what she said etc). This is why USB-C is so good. It can go in either way. I'm sure a user would still find some way to gently caress it up though.
|
|
#
¿
Mar 29, 2018 10:09
|
|
|
Those pictures don't make me annoyed or angry, they just make me very sad.
|
|
#
¿
Apr 3, 2018 10:16
|
|
|
Bob Morales posted:"Hey, can you give NON_MANAGEMENT_USER access to the file \\MANAGEMENT\SOMETHING\ELSE\REPORTS\BLAHBALH ?" My old job had something like this and because the requests came from director level and my boss was a bit dumb we ended up with the most hosed up folder structures where really important files and folders accessed by regular staff would be hidden behind management level only folders. It was like little parts of the folder structure were Switzerland that just did not give a poo poo about any of the permissions above it and operated by itself. So a huge amount of permissions on the folders went something like -> Read only -> Modify -> list folder contents -> list folder contents -> list folder contents -> FULL CONTROL. It was an utter shitshow and before I left I recommended that they just build a brand new folder structure for their data and force the users to migrate it themselves so they could at least have a semblance of logic behind their folder structure (nothing ever changed and is probably even worse today).
|
|
#
¿
Apr 10, 2018 14:39
|
|
|
Generally, non tech execs see what a proper DR plan costs and after finding out what it is for just scrap it (or severely neuter it) because "well we've never had that happen before  "And then they look great for saving the company a bunch of money. Of course the moment they realise that a DR plan needs to be enacted they are the first ones out the door swanning into a new job leaving everyone else to deal with the fallout. In the end, nothing is learned and the circle of life continues.
|
|
#
¿
Apr 12, 2018 10:09
|
|
|
When I was young I had an Atari ST with an attached tape drive. My favourite game was some 4-bit Bruce Lee thing. This must have been in 1986 or something. Edit: the tape drive took around 20 minutes to load a game. While it was loading you had to stare at some psychedelic flashing lights. Almost certainly would have triggered seizures in some people with hindsight. Dravs fucked around with this message at 15:39 on May 30, 2018 |
|
#
¿
May 30, 2018 15:36
|
|
|
CitizenKain posted:I don't know how people got through shooting that without losing it. Because their actors, they haven't lived through the actual horror of meetings like this. If I tried to do it I would probably break down crying halfway through.
|
|
#
¿
Jun 20, 2018 10:05
|
|
|
I was once asked to name a bunch of meeting room calendars in one of our sites after various Leicester City football players. Also, I once worked for an Opera company where all of the servers were named after various Operas which was especially confusing because they had no relationship to what the servers were even used for. Plus I know dick all about opera so that didn't help.
|
|
#
¿
Aug 29, 2018 11:31
|
|
|
Woo wooooooo woooooooooo Just got a really good opportunity in a senior Devops position for a huge company in the UK. Coming from purely Infrastructure Engineering so it's going to be a fun ride.
|
|
#
¿
Dec 18, 2018 16:42
|
|
|
When I was a bit younger there was a really nice pub on the edge of the forest with a big beer garden with a large lake behind. Periodically a gaggle of geese from the lake would just roll into the beer garden clearing it out because they are cunts. You could walk by the pub and see people cowering inside while the geese took over outside knocking poo poo over. How did this thread segue into geese?
|
|
#
¿
Dec 21, 2018 09:39
|
|
|
Farking Bastage posted:I just had a desktop tech basically drop a printer down on a table and send me a ticket to not only mac authenticate it on the network, but to " remotely set the IP and sleep settings." No, you lazy loving useless pile of gently caress. Do you want to know why you are a 50 year old desktop tech? This is why you are a 50 year old desktop tech. This kind of thing is mad to me. Where I am currently working there is a lady on 1st line who has been there for almost 16 years now. Absolutely no interest in pushing to get out of it. Just happy sitting there resetting passwords for close to 2 decades. I just don't understand the mentaility.
|
|
#
¿
Dec 28, 2018 10:02
|
|
|
Marcade posted:The display. She got the brightness and contrast so messed up she couldn't see anything. How she got there from trying to change the font is anyone's guess. It's like when a driver pushes the accelerator in traffic and instead of hitting the brake before their hit someone they panic and their brain just locks and they slam their foot down and smash into the person in front of them. Like that but with the contrast button?
|
|
#
¿
Jan 13, 2019 18:42
|
|
|
Weatherman posted:Is the veeam update process relatively easy? I have "install latest update" on my list of things to do but since I only learnt what veeam was last November, I'm a bit wary of changing in production. You literally run the update file and follow the wizard. I've also found it very resilient, the few times it has had an issue it has rolled itself back without a problem. Also, the Veeam support is pretty good, you ring up and the 1st person you talk to is the engineer who will be helping you. There is no talking to a call handler who assigns it a priority and promises a call back within 4 hours or so. This is in the UK at least, unsure if their support is different elsewhere?
|
|
#
¿
Jan 24, 2019 11:46
|
|
|
kensei posted:You're not wrong; we don't have any deployment of servers automated at this time. Do you at least have a virtual environment with templates you can roll out? Or are you literally going to have to build a bunch of physical servers, cos that would suck and I would say something like "welcome to 2005".
|
|
#
¿
Feb 5, 2019 11:21
|
|
|
Pyroclastic posted:unless a stock Intel HSF can get knocked out by the case falling over. And this is like the third time I've seen it over the past 4 years. Snap, push, and twist, and it's back to running at 3.2ghz. No, a correctly installed HSF would literally snap the motherboard in half before it became dislodged itself.
|
|
#
¿
Aug 22, 2019 11:31
|
|
|
Wow, some of these stories are triggering uncomfortable memories and remind me why I got out of production infrastructure and moved into DevOps.
|
|
#
¿
Jul 10, 2020 14:22
|
|
|
shortspecialbus posted:This doesn't absolve you of risk - my story earlier was puppet code that made a bad assumption about home directories being unique that resulted in significant irreplaceable data loss. Yeah but if I gently caress up the test environment at 5pm on a Friday I can just shrug and fix it on Monday morning. Plus my projects are all in Azure now, so no more having to do bullshit like power down tests or hardware replacement.
|
|
#
¿
Jul 12, 2020 10:49
|
|
|
|
| # ¿ Apr 25, 2024 06:24 |
|
|
larchesdanrew posted:After being stuck in a basement office for over a year day in and day out answering phones and struggling with the poo poo documentation of the last onboarding tech, this is going to be a dream 😅 Waiting for Act 3 where the old TV station becomes a client and you end up onboarding a bunch of buffalo NAS drives linked together with USB cables.
|
|
#
¿
Aug 20, 2021 13:44
|
|