DOH!
Dont ask me why but now I see the “Parallelism limited to 3 tentacles concurrently”
The UI however is showing activity on 6 items
Excuse the zoomed out image (Attached)
Attached is my raw output.
Parallel configs has been read as I can see these lines “Parallelism limited to 1 tentacle concurrently”
ServerTasks-8783.log.txt (108 KB)
Paul got back to me on email:
“
I think we misunderstood. MaxParallelism is designed for when you have a
single package going to many machines. In your case, you have different
packages going to each machine. We’ll make a change and put a new release
out shortly so that it works in your scenario. Thanks so much for sending
this through and sorry for all the trouble.
”
so hold tight, and we’ll be sorted!
The new release is out, with max-parallelism effective in the required way: http://octopusdeploy.com/downloads
Looking forward to closing this thread 
Thanks!
Thanks Nick! It’s working as advertised!
A small optimisation would be great in future however:
It seems to behave like this:
9 packages to upload.
It starts the first 3, and then it must complete those first 3 before doing
the next 3.
It would be sweet if it was always trying to upload 3 at a time. Otherwise
you wasting a lot of time waiting for the whole group of 3 to finish. It
leads me to wanting to order deployments in terms of package size so I dont
upload 2 small files with 1 massive one which would delay things.
MUCH APPRECIATED!!!
Thanks for the follow-up Anton, great to hear!
I’ll add your comments on the batching algorithm to a ticket for us to look at in the next release. Elsewhere in Octopus we use the true “rolling” approach you’re describing, but since there are a few layers of indirection in the package acquisition process it is a little more complicated to get right here. Definitely possible and desirable though.
Thanks again,
Nick
Thank you, it’s working!
Raoul
We’ve been seeing this problem on and off for the last couple of months, but just on one server/tentacle. Sometimes the upload of the 5 packages would be fine, but on other days they always fail with the “The message expired before it was picked up by the remote machine” error.
I thought maybe the server need a reboot, but a deploy immediately after gave the same error.
I’ve just set Octopus.Acquire.MaxParallelism to 2 and the deploy has finally worked which is great.
However, I suspect that it’s just masking the underlying problem which is ostensibly a slow network connection. I’m deploying from an Australian DC to a US east-coast DC, but both servers are configured with 100Mbps public bandwidth and neither of them show high network usage at the time. Maybe we’re just going via a bad network route?
Hi Paul, thanks for getting in touch.
Since this thread’s a bit stale/unwieldy, would it be possible for you to email support @ octopusdeploy.com with details of the Octopus version you’re using, and possibly OctopusServer.txt/Tentacle.txt log files (http://docs.octopusdeploy.com/display/OD/Log+files) from the time of one of these failures?
Best regards,
Nick