a delayed restart of instance happens. Change-Id: Iaabc2ffdccaa4c7c3bbfd8e974d7de52baa4d642
…fter configured number restarts fail."
configured number restarts fail. This augments exponentially delayed restarting of flapping instances. Also, adjust default values for parameters responsible for flapping behavior. Change-Id: Ie927040f276009e722c364fb50c1280acee6d47c
In some concurrent benchmarks `iptables` exited with a non-zero status and the message `Resource not available` (which I assume can be traced back to an EAGAIN in netlink I/O). Prevent this class of errors from happening by not executing `iptables` concurrently. Change-Id: I20b9cdd4397d79f16a1968b267b91a54b14cc6d3
This changes makes upstart run the jobs we need in a more predictable order. Because we use console.log to probe whether SSH has already started or not, we need to postpone starting rsyslog or it would capture everything written to /dev/console to /var/log/syslog inside the container. In addition, we want SSH to start when the virtual ethernet interface has been configured (or it still wouldn't be reachable). Change-Id: I20ec13e30eefa5e896797f5be17680ccacdbc0a4
…18f2,I3951d28c,Ia14b5b01 * changes: [Package Cache] Rakefile clean up [Staging] Rakefile clean up [Stager] Rakefile clean up [stager] Separate Gemspec [HM] Rakefile clean up [router] Rakefile clean up [DEA] Rakefile clean up [common][CI] Configurable Priority Queue Load
- `rake spec' no longer installs dependencies - no shelling out - single Rakefile - separate gemspec - replaced obsoleted gempackagetask with Gem::PackageTask - `rake install` continues to build and install stager gem Change-Id: If55229379ec28504f68d3f8d1bf92da3bf3c76c9