Since we added monitoring it became obvious that the mirrors.phx system is running out of memory often and swapping which causes unneeded I/O load on our storage servers. Memory should be bumped, alternatively we should see if we can somehow bring memory usage down, for example decrease number of jenkins worker threads (currently there's 5)

$ free -m
{{ total used free shared buff/cache available}}
Mem: 3788 3568 102 1 117 18
Swap: 3968 3968 0