New on LowEndTalk? Please Register and read our Community Rules.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
All new Registrations are manually reviewed and approved, so a short delay after registration may occur before your account becomes active.
Comments
I like that you take the time to add the notice but not to use another service/host your own server or similar. :')
I think (for me at least) it's because people - well, those who now something about tech and don't complain their 15GB of google drive is full and thus google should give them more storage for free - don't like to leech. You see, I am a person born after 1993 (so yeah @BruhGamer12, I'm a liar) and thus wouldn't have funds to have a seedbox running 24/7 when I usually only seed till I get somewhere around of a 1.2 ratio and then shut my home pc down again. I'd enjoy using a seedbox for sure, there's less stress, but eh, buying 9 cent seedboxes with new accounts or whatever is just dumb. But yeah, lastly, it sure isn't deniable that while piracy goes up again, torrenting/p2p in general/usenet etc. goes down.
We ran a similar service in the past, but these days? The amount of moderation to make sure absolutely nasty stuff doesn't get added ... Yeah.
Tho there would probably be a self hosted imgur clone out there, would be trivial these days to setup. Wasn't back in the day.
could just drop to some web server tho, but running a node just for hosting images to post to forums doesn't sound like efficient use of resources either
There's a ton of public direct file/image hosts. A lot of them you can upload to with just a curl command too.
0x0.st
catbox.moe
pomf.lain.la
...
They also integrate into a screenshotting software called ShareX.
imgbox allows hotlinking - https://imgbox.com/
Been online for a long time but sometimes the site is down multiple hours consecutively.
Yup. I used to run one as well but it did go invite only really fast. I did kind of mean a private service.
I totally get ya. Only makes sense if you've got an idler with some storage left or your own infrastructure. In that case I'd use the amazing and very professional service https://github.com/tycrek/ass or the ones mentioned by matey0. There are of course thousands more, but these do the job well.
some of those i have tested, they are slow.
and i consider them ephemeral
should just run our own i guess, would solve both issues.
Once you got your own infrastructure you realize how damn expensive everything is, and you start to avoid anything that might need maintenance like the plague.
Some True Stories:
Few years back, there was a brownout causing 1/3rd of our production to go down, mainly because of 1) Fucking Eaton PowerWare UPS 2) Buggy BIOS/BMC halting whole system. Ok, no biggies, just go click a button on the UPS, and manual power cycle the nodes down.
One gotcha? It's 31st of December, 23:58 and after right about 10-15 servings of magic social lubrication potion.
Sidenote: All the downed nodes were because of those 2 reasons, fraction of second brownouts PSU caps are more than enough to keep voltages up btw. If you get a real power outage, those always last longer than a UPS can handle anyways and you need genies at that point. Typical UPS is designed to bridge gap until genies are running, at speed and have some temperature in them. So approximately 98% of UPS installations are just psychological, not actually useful and actually quite the opposite of intended purpose.
Right now, a switch decided to start overheating last night, it worked for few hours, just enough for me to decide it's ok until next business day, i can go to a annual event friends were asking me to go to. Couple hours later, and the switch is rebooting again every ~10minutes or so, causing outages for approximately 100 users, so not that widespread. But instead of going to the event, i'm about to go replace the switch OR replace it's thermal paste and check all fans are ok.
It's pegged constantly at 91C, so thermal throttling too. It's mostly MD nodes, but also a few seedbox servers still on that switch.
You're the only person?
as the business owner shit rolls upwards. Yes, this kind of shit defies gravity.
It's easter holiday and i didn't request others to be on call during this holiday.
The joys of running a hosting business; You need to have 24/7/365 monitoring and be on call 24/7/365 -- you never get days off, ever. Even if i get someone else to be on call, as the business owner i'm always regardless on call, as whoever is on call might not know the specific thing they need to repair.
and for me personally, just the driving takes 2hrs.
-Aleksi
Only true for small to mid-sized businesses I think. I doubt Mr. Hetzner is on call 24/7/365
Nah, they probably are (at least for business people) - they just outsource it.
Yeah of course the company itself has to be available 24/7 but there is no one central person taking all the load and responsibility. They probably have like 5 top engineers who each are able to keep the shop running and know all the details.
Storytime!
But yeah, I know from own experience that having maintenance - "Oh, it's just a quick upgrade" - one hour later, I'm still trying to fix everything that broke.
Regardless, that does sound shitty. Especially since considering the effect I can't remember the name, it always happens on the dates where something important's happening.
Yep. At least probably.
I assure you, when that OVH DC burned, Oles was there and all the other founders / top execs.
Same goes for Hetzner, something bigger breaks, the CEO is dragged out of their vacation.
This is true for all companies. That's why they get paid the big bucks.
What changes, is the size of issues they have to be present at. Not for a simple small switch change. Then again, how often does a switch break? Not very often.
But someone like hetzner or OVH, say their router responsible for AMS-IX connections dies, yea the CEO is going to get pinged for that, as that's big enough to get some grievance from large business customers.
It can be as small as a fiber being cut, and all the top executives will be woken up. That happened to Elisa, maybe largest Finnish telco, but not quite Telia size as Elisa hasn't spread multinational. A single fiber was cut, and all the top executives had to start doing damage control.
We can't afford 24/7 on-site staff, so someone has to be on call at all times, that's typically me personally because i'm a bit stingy with money (rather buy servers than pay for on call duty). For holidays and weekends i typically only arrange someone else when i'm too far away to get to DC within sensible time.
"Fun" side note about Finnish bureaucracy; Because of our strong unions etc. to have true 24/7 coverage you don't need just 3 people (8hrs each/day) but more like 6, full time. It's 21 shifts per week, each employee can work for 5 shifts per week, add vacation time, extra downtime for 3 shift work, lunch breaks etc. and it drops to more like 3.5 shifts per week per employee, and since you get free sick leave for couple of days without doctor's notice, add some headroom for hangover sick leave and you get only like 3 shifts per week per employee and turns out you need 7 people just for "NOC".
Now add our insane taxation (and bureucratic costs), high cost of living, and the night work bonuses of +100% and you are left with minimum of about 50 000€/Month in salaries just to keep true 24/7/365 watch.
They could build servers tho at that time, provide customer support etc. ofc, so true extra is "only" about ~44 000€ per month over 8/5
Now to get that 50 000€ in profit ... You have to sell at least something like 500 000€ per month. Preferrably 2 000 000€ per month to have decent per head revenue, for IT industry.
So just to have true 24/7/365 monitoring and on-alert, completely by the book, we are already talking about 24Meur annual revenue business.
We are not there. Even after the new DC is built and filled with servers we don't expect to hit those numbers.
ah resolved, but now we get to other thing people don't realize / talk about;
Now that this networking issue is resolved, i have to sit here and watch things for a moment, with nothing tangible to do. Not much sense to start something which takes hours to complete or needs more thought.
Issue was kind of stupid -> Some of these switches have intake filters, somewhat coarse. But we are talking 40mm fans here, not 120mm or larger.
Even tho our DC is generally so clean that you could eat of the floor, this switch had accumulated enough dust in the intake filters to clog them up.
Regardless, only noticed this after prepping another entire switch and replaced the full unit.
soo ... i just spent hours on something what 30 seconds with vacuum would have solved and/or 5 minutes when setting up the switch originally (removing the filters). This is how little things stack up at a DC. It may just be a 0.5gr part, but if it's missing or in wrong place, you are going to have bad time.
[in monotone voice] and now back to regularly scheduled programming .... --->>
Has anyone tried to run Proxmox with a single VM and all (sans 8006) ports routed to the VM on the pulsedmedia MD platform?
Any hints regarding the network/interfaces?
There are probably better ways to do this. But what I did was setup a nat to route the traffic to their internal IPs and "port towarding" in the IP tables routing configuration. Just an option