Since run.sh currently does no more than start the node process in the way we used to before, this can easily be circumvented by invoking the agent old-style with node agent/vsoagent.
However, if run.sh ever should become a more complicated script, it should pass on command-line arguments to the agent
Previously (before we had run.sh) you started the VSO Agent by calling
node agent/vsoagent
.When doing that you could specify switches that configure the agent. E.g.
-u <alternate username>
The guide for On-Premises-deployment (https://github.com/Microsoft/vso-agent/blob/master/docs/onprem.md) states that you should invoke
./run.sh
to start the agent. The arguments specified torun.sh
are not propagated to the node process thatrun.sh
starts.Since
run.sh
currently does no more than start the node process in the way we used to before, this can easily be circumvented by invoking the agent old-style withnode agent/vsoagent
. However, ifrun.sh
ever should become a more complicated script, it should pass on command-line arguments to the agent