3 ms·

This is a very old problem, I used to see it decades ago when making tape backups. Tar would use move the entire disk through the buffer cache so that eventual

This is a very old problem, I used to see it decades ago when making tape backups. Tar would use move the entire disk through the buffer cache so that eventually everything in it was paged out. The classic solution was to use unbuffered versions of the disk device for backups. What I've always thought is that there should be a working set size limit on a process which includes the buffer cache somehow. The idea is that the process may not use more RAM than this size- if it exceeds it, it must either fail or swap out its own pages, not those from any other process. This would fix the problem for tar- it only needs a tiny amount of memory. I think the situation is very similar with the web-browser example. The browser should not be allowed to force all unrelated data to be paged out.

throwaway8941 7y ago
>working set size limit on a process which includes the buffer cache You can control this with cgroups. Plug a process into a separate cgroup and set the `memory.limit_in_bytes` knob to whatever your heart desires. I use it to limit the qBittorrent's memory usage on my machine. `firejail` is very convenient for doing this. If I don't set a limit (30% RAM in my case), it eats up all the memory with a uselessly large file cache, which does not improve upload speeds at all. https://access.redhat.com/documentation/en-us/red_hat_enterprise_linux/6/html/resource_management_guide/sec-memory https://access.redhat.com/documentation/en-us/red_hat_enterp...