Re:Lost 72 hours worth of research data to forced update
by nuisance-bear on Saturday April 10, 2021 @11:18PM
(#26767236)
I'd say this person is actually lucky Windows forced an update. Compute nodes fail, which is why you log state so you can restore your simulation or training routine. Instead of experiencing a failure in industry, they learned the "cache state" lesson in a low-stakes academic setting. Maybe they'll extrapolate the lesson and start keeping off-site backups too.
Re:Lost 72 hours worth of research data to forced update
by 4gotunameagain on Sunday April 11, 2021 @02:51AM
(#26768057)
Well this is not applicable to all scenarios/available hardware. We don't even know what the computation was and how powerful a machine he had access to
Re:Lost 72 hours worth of research data to forced update
by molyss on Sunday April 11, 2021 @04:10AM
(#26768491)
If you have enough ram to keep the current state of your computation, you probably have enough disk to push it to, several times over (for obvious reasons, you should never overwrite your latest snapshot with the most current data). If your computation was on a self-managed windows version, you definitely have enough disk.
Re:Lost 72 hours worth of research data to forced update
by jeppesen-io on Saturday April 10, 2021 @09:07PM
(#26766723)
Flash back to a few days ago when my Macbook rebooted and started the upgrade/patch process soon after a Zoom meeting ended. Lost 30min of work then went into a boot loop
Suprised me - I didn't know MacOS did that. Happened to my boss as well
Re:Lost 72 hours worth of research data to forced update
by ziml77 on Saturday April 10, 2021 @10:48PM
(#26767133)
I have the opposite problem with macOS. It tells me there's an update, I tell it to do the update later expecting it to take care of that overnight, come back the next day to another notification telling me the update wasn't applied.
Re:Lost 72 hours worth of research data to forced update
by smt88 on Sunday April 11, 2021 @12:13AM
(#26767432)
Assuming a process won't fail or a machine won't reboot for 3+ days is a bad idea on any OS, although it's certainly a worse idea on one that's targeted at the general public who only use Chrome and Office.
Writing the code in a way that can't be resumed is the person's second mistake.
Given this anecdote, I still prefer auto-updating Windows because it keeps everyone more safe. It's terrifying that there are millions of machines running Windows with access to my personal financial and medical data, and many of those may have 0-days on them. I don't know how we got by when those devices sometimes weren't fully updated for yeads.
Re:Lost 72 hours worth of research data to forced update
by passivate on Saturday April 10, 2021 @11:45PM
(#26767332)
You need to tell IT to use WSUS. I've managed to convince IT to stick one my test stations into a 'never update' bucket. Months and months of uptime! :)
Re:Lost 72 hours worth of research data to forced update
by detaro on Saturday April 10, 2021 @11:22PM
(#26767247)
at least in my experience, setting it through group policy so that it will not download patches without user interaction works. without downloaded patches it also won't install any automatically.
Re:Lost 72 hours worth of research data to forced update
by kwyjobojoe on Sunday April 11, 2021 @03:59AM
(#26768441)
It's not exactly hard to prevent windows 10 from running updates. My security cam server has been running for years with updates only applying when I manually run them.
I do wish that Microsoft made it easier to prevent auto rebooting updates, but you can do it and it's pretty stupid to run a multi-day process without bothering to set up the system correctly and then bitch about it.
Re:Lost 72 hours worth of research data to forced update
by ziml77 on Saturday April 10, 2021 @10:40PM
(#26767104)
I don't recall ever being forced to restart Firefox. It just puts a message in the menu that says "Restart to update Firefox". But I can let that sit there as long as I want.
EDIT: Actually just looked it up and I see how you could have the issue. If you're using your package manager to update instead of using Mozilla's updater, then it will perform the update without accounting for whether or not Firefox is running. That means that the next time Firefox loads something from disk, it will mismatch with what the running version expects.
But of course I agree, if you can, you should
Suprised me - I didn't know MacOS did that. Happened to my boss as well
Writing the code in a way that can't be resumed is the person's second mistake.
Given this anecdote, I still prefer auto-updating Windows because it keeps everyone more safe. It's terrifying that there are millions of machines running Windows with access to my personal financial and medical data, and many of those may have 0-days on them. I don't know how we got by when those devices sometimes weren't fully updated for yeads.
I still haven't figured out which (if any) group policy settings will prevent automatic reboots. Hours of my life wasted.
I do wish that Microsoft made it easier to prevent auto rebooting updates, but you can do it and it's pretty stupid to run a multi-day process without bothering to set up the system correctly and then bitch about it.
EDIT: Actually just looked it up and I see how you could have the issue. If you're using your package manager to update instead of using Mozilla's updater, then it will perform the update without accounting for whether or not Firefox is running. That means that the next time Firefox loads something from disk, it will mismatch with what the running version expects.