Sorry to resurrect an older thread.
We're currently on 8.6 RU3 and still seeing this exact issue - it doesn't always happen, but when it does, it creates major complications. Although this has been tolerable for the last few years, we've seen a spike in this behaviour over the last week, with roughly 30% of builds stalling at Boot To Production. They do not correlate to site, operating system, or hardware.
The workaround for us is similar to Greg's description - we log in and hit the Reset Agent button in the Agent. The machine then resumes the deployment sequence without further difficulty. Whilst this works, it does defeat the purpose of automated deployments.
We run a dedicated Task Server alongside the primary Altiris server - I remember seeing discussion elsewhere, possibly in this Community (I've long since lost the link) about configuring IIS on the TS to use only HTTP connections, which we did during the initial diagnostic and mitigation a few years ago. This helped reduce the frequency somewhat but did not eradicate the problem entirely.
I am keen to hear from anyone who has either fully resolved this problem in their environments, or, is still seeing this issue on 8.6 RU3.
Thanks all