Timeouts and retries when publishing artifacts

Summary

Recently, some Cloud instances are failing to publish artifact with an error. Other bugs when publishing bugs happened with different errors.
Note: the bug is not limited to Cloud, it was spotted there because it became more frequent with the recent update.

Steps to Reproduce

1. Try to run a plan with an artifact

Expected Results

The artifacts are published.

Actual Results

The below exception is thrown in the Bamboo logs:

2015-10-06 06:54:31,790 INFO [0-BAM::Elastic Agent on i-12345678::Agent:pool-3-thread-1] [RetryExec] I/O exception (java.net.SocketTimeoutException) caught when processing request to {}->httpt://127.0.0.1:46593: Read timed out
2015-10-06 06:54:31,790 INFO [0-BAM::Elastic Agent on i-12345678::Agent:pool-3-thread-1] [RetryExec] Retrying request to {}->httpt://127.0.0.1:46593

Workaround

For Linux Elastic Instances only
Add this to the Startup Script:

sed -i 's@.*MARKER.*@sed -i "s/-Xmx256m/-Xmx256m -Dbamboo.agent.http.client.timeout=3600/" $startupScript@' /opt/bamboo-elastic-agent/bin/bamboo-elastic-agent

For remote agents
add

-Dbamboo.agent.http.client.timeout=3600

to the agent command line.

For Windows elastic instances
None so far