Would be great if through the monitor, we could easily trigger a version update - which would ensure all related services have stopped running, and no new services are started before the update concludes. The same way we can do restarts etc.
Sadly restarting the slave with multiple slaves running on a machine only updates about 40% of the time.
The restart slave remote command should accomplish what you’re trying to do. If you’re seeing it fail to auto-update when Automatic Upgrades / Downgrades is enabled on the repository then that is a bug. Can you provide a slave log from a slave starting up after failing to update? Also, could you provide the launcher log from that machine?
Attached some failed update logs from two machines, hope this helps.
It usually seems to be dlls in use, so it can’t overwrite them.
d8updateLogs.zip (139 KB)
Thanks for the logs, we will look into this.
Thought of another use case where the remote update command would be useful… we have machines that are either offline or disabled (for a reason). In these cases, we can’t update them, since the slave should not actually be running on the boxes. But we don’t want them to skip the updates, as the machines can be used for submissions.
We’ll add this to the wishlist.