3 ms·
I run an Arch Linux mirror. [1] The server hosting the mirror is averaging 1 MiB/s (8 Mib/s) outgoing traffic. There's quite a bit of fluctuation since package
by majewsky 5y ago
I run an Arch Linux mirror. [1] The server hosting the mirror is averaging 1 MiB/s (8 Mib/s) outgoing traffic. There's quite a bit of fluctuation since package downloads are a very spotty load, but overall I don't think the server is breaking a sweat at all. CPU usage is pretty flat at 5% (of 1 vCPU). The same server is also hosting approx. 10 other websites, but the mirror causes most of the load by far from what I can tell.
[1] This is n=1 data because I can only speak for my own machine, so obviously take this with a grain of salt.
- bee_rider 5y agoHmm. So assuming the whole 250GB he mentioned was downloaded (I guess there must be some decompressing and automatic file generation going on, so probably not, but this is not a very accurate computation I'm doing), that's in the general ballpark of 10% of month's bandwidth usage?
- majewsky 5y agoIt's 250 GB uncompressed. $ df -h /var/lib/archlinux-mirror Filesystem Size Used Avail Use% Mounted on /dev/sdb 89G 70G 15G 84% /var/lib/archlinux-mirror Some of that is ISOs and the `testing` repo, so the actual download size will be in the ballpark of 50-60 GB, which would put it at more like 2%.