<html><head></head><body><div>On Fri, 2015-09-18 at 11:17 -0400, Jim Kinney wrote:</div><blockquote type="cite"><p dir="ltr">User was trying to transfer data to/from oracle on Linux. The connection kept dropping. The Linux system is a vm on ESXi. The physical host is a tad overloaded on memory not complaining about cpu cycles.</p>
<p dir="ltr">By telling oracle to use /dev/urandom instead of /dev/random, the connection drops went away.</p>
<p dir="ltr">Could the vm be not getting enough cpu cycles to keep entropy pool full or should I look somewhere else? Ideas to boost entropy collection rate are appreciated.</p></blockquote><div><br></div><div>When using the QEMU/KVM stack, the solution is to have a really strong HWRNG on the host, and pass through some of that entropy to the guests. I don't know if there is a way to do that on the stack that you're asking about, though.</div><div><br></div><div>If you have a Rπ or Rπ2 laying around, install a barebones buildroot with rngd, openssh, and a set of ssh keys, and use its HWRNG to feed the VM. The HWRNG will generate random data far faster than a typical desktop or server will. (The /dev/random device then responds more like the /dev/urandom one does, but with more entropy.)</div></body></html>