Running logrotate -d on cluster with a lot of slaves results with execution
taking a lot of time (>10s, one some nodes ~20s), which leads to random
promise timeouts.