[why] when run multi-docker job on physical machine, run 50 client to request job, if server down or error like Connection refused. these client will exit. when the server recovers, need to run these client again.
[how] add exception handing to ensure all client does not exit --- providers/docker/docker.rb | 18 ++++++++++++++---- 1 file changed, 14 insertions(+), 4 deletions(-)
diff --git a/providers/docker/docker.rb b/providers/docker/docker.rb index c241d3c..2239930 100755 --- a/providers/docker/docker.rb +++ b/providers/docker/docker.rb @@ -88,6 +88,19 @@ def main(hostname) run(hostname, load_path, hash) end
+def loop_main(hostname) + loop do + begin + main(hostname) + rescue StandardError => e + puts e.message + sleep 10 + ensure + sleep 5 + end + end +end + def save_pid(pids) FileUtils.cd("#{ENV['CCI_SRC']}/providers") f = File.new('dc.pid', 'a') @@ -99,10 +112,7 @@ def multi_docker(hostname, nr_container) pids = [] nr_container.to_i.times do |i| pid = Process.fork do - loop do - main("#{hostname}-#{i}") - sleep 5 - end + loop_main("#{hostname}-#{i}") end pids << pid end