Intermittent EOF traces w qemu vms

[jamesmarkchan]

During @jslcom's tests of 0.6.3-dev.fat the rpm distribution in qemu, intermittent stack traces were observed on fedora linux 24 workstation. These did not occur in the local NUC hw test with the fedora ws.

these have been reproduced in a debian qemu vm, potentially a resource race condition but it would be good to understand the issue and handle these stack traces appropriately. It is believed theses have been isolated to qemu vms.

This is the line throwing the EOF exception:

This behavior only occurs with benchmarks w larger sample sizes above 50.

Fine up to 50, five exceptions at 100. Then two [exceptions] at 170. Then none.

[jamesmarkchan]

would be good to know if this is only with read operations