You are not logged in.
Pages: 1
I guess your web server creates new processes for each incoming connection. When it starts a new process, it might get an incomplete binary if it's being updated live. I guess that might be what happens, although whether you'd classify that as a kernel oops is up to you. I get a seg fault when I cropped a sample binary just now at almost any length, but ymmv.
I'm not sure the best way to resolve this. I guess big companies would use two servers and flip flop between them. You could also take the http server down while you're doing it I suppose, and given we usually have multiple addresses in our mirrorlists that should resolve to someone else unless everyone's updating at exactly the same time.
Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.
Offline
I don't have a Arduino-Raley-Reboot-Thingy (yet) and I'm not physically close to the machine, otherwise
a hard reset would not be an issue. Hosting on >10 year old hardware is maybe not the best choice either.
With kernel OOPS I meant a kernel panic when executing 'pacman -Syyu'. In no part of my universe this
makes any sense.. :-)
Offline
I think everyone using arch32 that needs to must be using >10y old kit at present. But it's great news that it's all back and tickety boo.
Architecture: pentium4, Testing repos: Yes, Hardware: EeePC 901+2GB RAM+OS half on the SD card.
Offline
Now it was my web proxy (Raspberry pi). I'm currently getting so many attacks on the mail server on the same machine, crazy.
I'm refraining from making my whole personal infrastructure redundant... for now. :-)
Offline
Pages: 1