If you feel strongly about it, you can open a bug report and even contribute the improvement to the protocol. I'm sure the Jenkins committers will welcome the contribution.
And if java coding isn't your cup of tea, you can also help others by adding a page to Jenkins wiki about this. -- Sami Rich Pixley <rich.pix...@palm.com> kirjoitti 31.7.2012 kello 0.12: > I do. > > The protocol should have a more robust initial negotiation in order to cover > exactly this extremely common case. Even reversing the initial protocol > message so that it comes from the slave would probably do the job. > > --rich > > On 7/30/12 11:50 , Sami Tikka wrote: >> Doh! Of course! >> >> I don't really think it is Jenkins' fault the earlier ssh clients ate the >> first protocol messages. >> >> -- Sami >> >> Rich Pixley <rich.pix...@palm.com> kirjoitti 30.7.2012 kello 21.40: >> >>> I've gotten this working. The clue for me was here: >>> http://jenkins.361315.n4.nabble.com/Doing-something-else-before-invoking-slave-jar-in-slave-launch-script-td4634608.html >>> >>> There appears to be a bug in the jenkins protocol but adding -n to the >>> earlier ssh commands in my runslave.sh script appears to work around the >>> bug. >>> >>> --rich
