how did we get back to linux having to be rebooted as often as windows
Timeline
Post
Remote status
Replies
16
@sun (this is status.poast.org btw probably the most stable shit we have ever lmfao)
@sun web browsers and nvivia drivers
Killing web browsers on hibernate will get you a week of uptime; the nvivia driver never integrates with the kernal on update s you if you add it to your ignore list you start getting months of ptime
Killing web browsers on hibernate will get you a week of uptime; the nvivia driver never integrates with the kernal on update s you if you add it to your ignore list you start getting months of ptime
@hakui I updated the server software not pleroma!
@graf do you livepatch the kernel
@sun All of our ubuntu hardware is livepatch yah
@graf incidentally the reason I had to reboot so many times was because I was turning on iommu in the kernel and twiddliing switches to blacklist nvidia so I could pci-passthrough my Tesla P4 to a VM
@sun I genuinely don't understand how you've had so much issue with it ours just worked from the start. Although I had to rewrite the Nvidia transcoding plugin for peertube, it works great. I am admittedly not passing through to a VM but that should be trivial I would think
@graf supposedly it all comes down to the hardware. the first machine we put it in didn't work at all, this machine works fine. I just wanted to isolate it into a VM so I didn't fuck up my host config
@graf passthrough worked, I just didn't know exactly how to do it. it took AI about 5 tries to get it right
@sun Like two weeks ago I tried to use chatgpt and gemini to fix something in a language I am not very familiar with and it hallucinated from the onset. What are you using? Claude? I get perplexity for free from Poast's phone carrier I should try that
@sun @A3J2ULq1MEFScUVADA.graf@poa.st i love claude with dangerous mode running in docker. it can be exactly as bad at shit as i am, but is much faster
@graf claude and chatgpt.