AWS Thinkbox Discussion Forums

deadlinecommand does not properly resolve slave host

When running deadlinecommand -ConnectToSlaveLog (or, I imagine, any other command that takes a slave name), the slave name is not properly translated into a fully-qualified domain name.

We have two repositories with a large number of duplicate slave names between the two. If I’m viewing the queue for the remote repository (that is, the one under a different domain), the ConnectToSlaveLog feature just connects to the local slave with the same name (or fails if there isn’t one).

I’m assuming you’re already using the “Use Fully-Qualified domain name […]” option in Repo Options -> Slave Settings? Does the fully-qualified name properly show up in the ‘Machine Name’ column of the Monitor’s Slave list?

I’m mainly wondering if this is a problem with us not managing to get the FQDN in the first place, or if we’re just not using the proper value in DeadlineCommand…

I tried with and without this setting enabled (restarting the slave after changes).

Yes, regardless of the state of the aforementioned repository setting.

Any chance this is fixed in 6.2.1 and/or 7.0? It’s makes working with remote repositories kind of painful…

Seems like this bug report got lost in the shuffle. Sorry about that.

When you launch the Monitor, are you using the -repository command line option to connect to the remote repository? If so, I think that would explain the problem because the NetworkRoot property in your deadline.ini file would be pointing to the original (aka: local) repository, and the ConnectToSlaveLog command currently connects to the Repository defined by the NetworkRoot property

Let us know if this is the case, and we’ll log it as a bug for 7.0.

Thanks!
Ryan

Yup.

It does appear that if I go to “File -> Change Repository…” and then point it at the same repository, this starts working properly.

Thanks for confirming! We’ve logged this as a bug targeting 7.0.

Cheers,
Ryan

Privacy | Site terms | Cookie preferences