I want to slow my outgoing traffic when it encounters congestion, and measure "congestion" by packet response times. The intent is to avoid one NFS client starving all the other clients when performing large writes.
I've read a good deal of manpages, tutorials, and serverfault answers related to tc traffic throttling, but they all generally start with picking an arbitrary bandwidth cap, and i'd really like the system to automatically adjust itself when our networking components improve.
I'd like a more systematic way for prioritizing my tasks with nice(1).
Obviously 0 is "normal", 19 is "lowest priority" and -19 is "highest priority", but what about all the other numbers?
If I have a job that is non-interactive (I'll go get coffee while it runs), I'd like to lower the priority a bit, but to which number? Should it be 1 or 2 or 10 or what?
Are there any "well-known" things that each priority number is used for?
The same questions apply to ionice.
At my old company we used NetApp, which supported a ".snapshot" feature. Whenever I did something stupid to my files, I could cd .snapshot and find hourly and nightly directories which contain snapshots of my current directory. I get the file I need and resume my work.
Is there anything similar in the open source world?
I find many backup projects, but they generally backup to a different machine, with no easy access for the user. I find a couple user-space filesystems which operate on top of rdiff-backup, but is there a project that ties everything together into a cohesive system?
Thanks in advance.
All my boxes have the same username for myself, except one legacy machine where I'm loath to try to change it. Whenever I ssh to the machine, I forget to specify my user name, and my password is rejected. This only takes maybe 30 seconds, but it should only take 3, so it's a PIA.
Is there any way to configure that machine to recognize the incoming username and map it to another, correct, one?
I'd really like to configure this on the server itself if possible. Configuring each client is less ideal, especially since I like to muck with my various installations and VM's quite a bit.