|
Don't forget their "oh what you REALLY meant to search for was this.." results if you enter a model number that doesn't exist as typed, but there's stuff in the database that's sort of close. Just tell me "nothing was found" you jerks.
|
#
¿
Feb 4, 2016 21:48
|
|
|
|
| # ¿ May 2, 2024 01:15 |
|
|
In practice it doesn't seem to be all that big a risk, we buy nexsan's high density stuff like it's going out of style and haven't lost a raid to drive failure in ~10 years. Couple losses from operator error though.  We did switch to raid 6 once drives got over 2tb, but there was no data protection motivation to it. We just had so much drat disk space it became "hell, why not?"
|
|
#
¿
Feb 9, 2016 01:41
|
|
|
Pfft, 12. Try 60.  (everything unproducible is on tape, spinning disk is just cache for faster access)
|
|
#
¿
Feb 9, 2016 04:14
|
|
|
Skandranon posted:That sounds insane. Backblaze only does 3 mirrors by 15xZ arrays (I think). Is this just a backup array that is barely referred to? Well they claim they use it full bore 24/7, but it's probably more like 75% of the advertised maximum throughput. It's high energy physics analysis, and the disks feed a batch cluster. Jobs are typically waiting for subsets of the data to copy to the batch node, and that's the least efficient part of the chain. Don't get me wrong, keeping it all going isn't easy.. swapping out disks is basically a full time job. But like I said previously in ~10 years I can't recall losing a single array to disk failures. And I'm not saying there's no risk and everyone should abandon best practice.. just that losing an array during a rebuild is extremely rare.
|
|
#
¿
Feb 9, 2016 17:11
|
|
|
Wonder what the conversation will be like when the office imbecile gets their account owned anyways.
|
|
#
¿
Feb 10, 2016 16:57
|
|
|
How about they get to keep the commission, but if delivery misses deadline because they promised features that couldn't be implemented, they lose their job. See if that keeps their mouth shut a little bit. World would be better in general, clients would benefit by not wasting money on vaporware.
|
|
#
¿
Feb 11, 2016 04:27
|
|
|
Manslaughter posted:Send her the same document but in the email say "here is a revised version with less technical jargon". She probably didn't even look at it. Used to work with an individual that was incapable of reading anything except the first line of an email. If you could cram all your instructions into that one line things were golden. If not? The only way to move forward was to head over to their desk and stand over their shoulder telling them what to do, step by step. Took me a little while to figure out, early on my emails were detailed and long, and I kept simplifying as I realized there was no point to it. The last straw was when I sent a message their query in the form of "yeah this is an easy thing to fix, here's the steps:\n\n1) <instructions>" and I got a response a few minutes later asking for clarification on what to do. From that point on it became a game to me, see how many instructions I could pack in those first 80 characters.
|
|
#
¿
Feb 12, 2016 17:49
|
|
|
Pfft, she's family. Anyone who does IT professionally is also the family IT support. "Oh sonny you know computers, well can you fix my printer??"
|
|
#
¿
Feb 12, 2016 19:11
|
|
|
luminalflux posted:There's a fair big difference in situational awareness in a factory environment with music from a speaker, earplugs/earpro and music in earbuds. Yep. Earplugs just muffle noises, you can still hear most everything with them in. They exist so you don't destroy your hearing. Headphones muffle the environment AND pump new noise into your ears, creating isolation from the world around you. They exist so you can hear a noise and no one else can. I can't imagine headphones in an industrial setting being a safe option for anyone.
|
|
#
¿
Feb 16, 2016 20:01
|
|
|
The poor sod hired to maintain our sharepoint sits near me in the cubicle farm, their desk is an endless stream of confused assistants and middle management cruising by to ask how long it would take to implement Impractical Feature #5383. The rare occasion that's not going on, I hear a lot of sighs and cursing.
|
|
#
¿
Feb 18, 2016 00:27
|
|
|
Developers can be super superstitious, you'd think being programming geniuses would mean they understand the computers they depend on a whole lot better, but nope. My anecdote is this one dude who insisted I reboot the company's build server after every single make, because the "top" command showed that not all memory was free and he insisted on his compile having access to every bit of ram it could get access to. I must have explained dozens of times how "used" memory was probably pages the kernel was holding on to for allocating to processes, but he never did figure out. Unfortunately the dude was untouchable because he'd written 75% of the company's product so I couldn't straight up tell him to gently caress off, I just had to apologize to everyone else in the company every time the server went down.
|
|
#
¿
Feb 19, 2016 17:39
|
|
|
Thanks Ants posted:
Wouldn't be surprised, the network interfaces HP puts into those things were hilariously insecure well into the 2000's.. long after every admin on the planet started to wake up on the concept of network security. For a while you could make any config change you wanted by formatting a message with a couple hex codes and firing it at the printer's IP address. It blindly did whatever you told it. I used to put joke status messages on the LCD screen of every printer on our floor, see if I could get people talking about the weird poo poo the printer was saying.
|
|
#
¿
Feb 19, 2016 23:10
|
|
does it talk entirely with broadcast traffic or something? How do you gently caress something up like that in TYOOL 2016.
|
Inspector_666 posted:Hosts don't communicate across subnets in any case, that's on the router, isn't it? Hosts need to know their gateway if they want to get past the local segment/router, and they need to know their subnet so they properly route remote traffic through the gateway. They sort of need to know their subnet if they want to talk to everyone on the same local subnet.. if everyone else has a /22 and one host has a /24, then it'll have issues communicating with any hosts beyond the first 255 addresses. What it sounds like to me, if the HP literally cannot talk to anything beyond the local network, is they just didn't provide a way to specify a gateway. Which would be colossally stupid, but as I said previously, the network stack on HP printers has historically been poo poo so I wouldn't be surprised if that were the case.
|
|
#
¿
Feb 20, 2016 06:37
|
|
|
Inspector_666 posted:Right but if the HP device can't accept a subnet mask or gateway address then how the gently caress is it supposed to communicate with anything? You can't give a thing just an IP address. If it can't accept a subnet mask, it's literally useless. If it can't accept a gateway it is only useless beyond the LAN. I guess if they hardcoded the mask to something, like 255.0.0.0, it might work for most local networks. But some engineer should get punched if that actually happened.
|
|
#
¿
Feb 20, 2016 06:43
|
|
|
Well when you make jokes about being a computer janitor, it's just a matter of time before someone takes you seriously.
|
|
#
¿
Feb 22, 2016 23:40
|
|
|
Collateral Damage posted:I'm honestly surprised MapQuest is still around. It's actually not horrible anymore, as near as I can tell it's feature equal to google maps. It just gets ignored because everyone defaults to google.
|
|
#
¿
Feb 23, 2016 17:33
|
|
|
Well in a sane environment the first incident is "free" to get the place to a point where they're treading water, followed up by a month of debriefings and root cause analysis before someone comes up with a scheme to prevent it from ever happening again. Usually by swinging all the way to the other extreme, requiring so much redundancy and red tape that no one can actually get anything done.
|
|
#
¿
Feb 23, 2016 23:22
|
|
|
Sickening posted:"Hey IT, we just bought this 20k printer and we need help getting it configured for our network." "Don't worry about the cable run, its already taken care of." I like the convenient foot-catching loop pre-configured to trip someone and break their neck.
|
|
#
¿
Feb 24, 2016 21:51
|
|
|
Downside is the world is always creating new idiots so even when the boomers are gone not much will change.
|
|
#
¿
Feb 26, 2016 02:17
|
|
|
Pfft, slackware. Give 'em a stack of floppies and point them at the linux howto archive and tell them they can't leave the house until they have a PPP connection working. Gentoo on the other hand.. that is in fact child abuse.
|
|
#
¿
Feb 26, 2016 16:11
|
|
|
My threshold is if I have to do a repetitive thing more than 10 times or it's something I have to do at regular intervals.. it gets turned into a script.
|
|
#
¿
Feb 29, 2016 21:58
|
|
|
"Oh hey everyone there's some large files I need to purge from our git repo, so I'm going to have to ask you to delete your entire tree and reinitialize. Please do not commit any of your pending changes or it will mess this up, oh yeah I've already done this so if you had a work day's worth of commits to push well tough poo poo, have fun merging them in manually." You couldn't have given an hour warning for this?  edit - based on my moderate knowledge of git merely purging files from the repo shouldn't require a complete re-initialization, but trying to do a pull on the tree generated conflicts for hundreds of files I had never modified so who knows what the gently caress they were actually doing. xzzy fucked around with this message at 23:59 on Mar 2, 2016 |
|
#
¿
Mar 2, 2016 23:32
|
|
|
3000 with 4 is doable.. if all the machines are identical. It's when people start fragmenting chunks of nodes into unique snowflakes that poo poo goes south. And being on call every day of the year is a show stopper.. +$25k or +$50k, I don't care. People deserve to be able to go out at night sometimes. 
|
|
#
¿
Mar 4, 2016 16:10
|
|
|
I'd consider it for seven figures, if I was younger and single. Live a lovely life for two years then buy a cabin in the mountains and never work again.
|
|
#
¿
Mar 4, 2016 17:27
|
|
|
Actually planes do have to constantly adjust their pitch, that's why they're all required to have trim controls so the pilot doesn't get exhausted trying to fly level.
|
|
#
¿
Mar 4, 2016 22:45
|
|
|
And the mechanics they implement in their theoretical round world behave suspiciously similar to real world rocket mechanics.
|
|
#
¿
Mar 5, 2016 18:19
|
|
|
Escape trajectory would be pretty simple, just roll off the edge. Probably crash the program trying to simulate what happens next though. It's nothing but NaN down there.
|
|
#
¿
Mar 7, 2016 21:37
|
|
|
Or bring a laptop and wheel over a chair and sit outside his office until he shows up. Master looking like you've been there for an eternity. Passive aggressiveness works!
|
|
#
¿
Mar 8, 2016 20:41
|
|
|
Yeah, I'm never taking care of a co-worker's kids. gently caress that, I'm out.
|
|
#
¿
Mar 9, 2016 16:44
|
|
|
My home system has realtek as well and it's been fine with whatever drivers W10 installs by default. Realtek ALC887, driver version is 6.0.1.7535. I would assume a hardware fault before blaming software, or maybe a loose connection somewhere. But "stuttering" can mean a lot of things. What does bug me about W10 is hibernation is still lovely. It's about a 50/50 shot whether the system will wake up when you tap the space bar.
|
|
#
¿
Mar 15, 2016 15:19
|
|
|
Yeah but if the angry idiot is a salesman, there's a high chance he's untouchable and nothing will actually come of it.
|
|
#
¿
Mar 18, 2016 02:06
|
|
|
Potato Salad posted:Halfheartedly pissing me a little bit off: My latest security related annoyance is the phone app for my bank. If you don't remember to hit the "log off" button every time you finish with the app and your login expires, the next time you open the app you get an alert popup about how it needs to "securely log you off." It can't just do this basic task, it obligates you to press "ok" every single time it happens before you can do anything else with the phone. I don't even know what "securely logged off" means. If the credential has expired, I'm already logged off, so just purge your cache and give me a login prompt. Morons.
|
|
#
¿
Mar 21, 2016 17:27
|
|
|
Judge Schnoopy posted:Legacy. Turns out it's really hard to fire a 20 year employee because IT wanted to install new software that proved 'difficult to learn', regardless of how easy it actually is. Plus ageism is a thing these days, and if some angry boss is all "we gotta fire him because can't/won't learn anything new" well then mr. old fart has ammunition for a lawsuit. It works both ways too, there's a lot of "this wet behind the ears kid keeps loving up because he's learning as he goes, punt him to the kerb" out there. Basically they gotta be cautious and follow some procedure (presumably cleared by the legal department) to avoid problems down the line.. build a paper trail to document that they can't perform their assigned duties, violation of office rules, stuff like that.
|
|
#
¿
Mar 21, 2016 23:29
|
|
|
Public schools should teach everything WAY better. Everyone's all for it, until it comes time to pay the bill.
|
|
#
¿
Mar 22, 2016 16:33
|
|
|
To be fair, I've racked a few thousand machines in my day and sometimes a good bang and a shove is what works. It's very rare to get rails mounted perfectly level and machines tend to rub against each other a lot. Though I tend to be aware of the signs that something's wrong.. if it's a hard stop or you feel something flexing, that's a good sign you done hosed up. It doesn't help that rails are always complete poo poo. They change the design with the seasons so the chances of getting the same type between purchases is super small, so you're always re-learning how to rack stuff.
|
|
#
¿
Mar 22, 2016 17:54
|
|
|
pixaal posted:Partly due to weight, but I've always been told to take all disks out before mounting something with more then 4 drives. Yes, it's a pretty good idea once you get over 10 drives. That weight adds up fast. Though I have moved 24 bay servers with a buddy, it's not fun. Once you get into the 40+ bay territory you have to take the drives out to avoid the rails failing.
|
|
#
¿
Mar 22, 2016 18:04
|
|
|
As long as all the drives in the unit are on the same controller, which bay they get inserted to isn't relevant. Well, as long as the controller saves unit information on the drives and can rebuild the config from that. Though I still try to keep the ordering the same, mostly to appease the IT gods and prove that I am detail oriented.
|
|
#
¿
Mar 22, 2016 18:36
|
|
|
Jeoh posted:I don't really get how mixing up the disks would break the array. I mean, they have unique identifiers, right? I never actually tried it myself, but I was warned back in the SCSI and PATA days that raid controllers could not stripe a unit across interfaces. The person that informed me of this was way more senior so I just accepted it as truth and never tried to explore it myself. Either way, it's a non-issue now as controllers are better featured and more powerful. Though now you have the problem with extremely large servers carrying more than one controller, so you at least need to make sure the disk is attached to the same controller it used to be.
|
|
#
¿
Mar 22, 2016 20:37
|
|
|
Paladine_PSoT posted:poo poo pissing my inner pedant off: XEON I don't know, I'd love to give certain users xenon based computers, because it's useful as an anesthetic.
|
|
#
¿
Mar 23, 2016 21:34
|
|
|
|
| # ¿ May 2, 2024 01:15 |
|
|
All distributions are poo poo in their own unique way.
|
|
#
¿
Mar 25, 2016 19:48
|
|