AWS Thinkbox Discussion Forums

Alpha 4 Feedback

Here’s some feedback mainly on the fancy new features in alpha 4:

  1. re: ping slave user permissions. As power user with default permission to “ping slave”, if I right-click on 1 slave instance, the ping command fails as the other slave instance has not been selected as well.

  2. re: ping slave permissions. Apart from the issue above, the new user permission seems ot be working as expected with 1 exception. The ‘new’ ping icon in the slave list is still able to be used. I was under the impression this would also get enabled/disabled depending on the user permission setting as per “ping slave” function?

  3. re: right-click launcher, “Launch slave by Name” - how about a slave’ type icon for this new function? In fatc, a few icons could be made for a couple of the functions in the launcher-right-click menu system.

  4. re: “Launch Slave at Startup” - which slave(s) will it start? Just the 1? or ALL slaves on that particular “machine name”?

  5. I was able to create a new slave with the slave name = “,” which crashes when you try and remote start it up. Need a mechanism to stop illegal chracters being entered into the new slave name dialog field. Ideally, I would like to ban everything except alphanumeric, which I think the label states this, but doesn’t actually enforce this? Also, I would love to be able to control the slave name that gets created. ie: the same as the original slave name BUT with a “_01” assigned to the end of the machine name. Going forward, I think this is going to be really useful, so that wranglers/studios can have a repository option to help them control the naming convention of slaves BEFORE it gets out of control and hard to identify things quickly in the queue. Yes, I know you can use the actual “machine name” to work out what is happening. Just thinking about usability on a day-to-day basis.

  6. After you have created a 2nd slave for the first time, in the launcher right-click menu, under “Launch Slave by Name” item, ONLY the latest newly created Slave Name is present. IE: The original computer name, slave is not shown. However, if I then delete this 2nd new instance slave name, the 1 original computer name based slave name re-appears. Seems wrong to me.

  7. Auto Configure Pulse is a nice touch. Works well. Just need the ability to have reduandant Pulse’s now, which are aslo self -healing if any part of the web server is detected to not be responding in a timely fashion :slight_smile:

  8. Deadline submitter interface within Nuke needs to be a little bit taller to remove the vertical scroll bar when you first open the interface.

  9. Q. Does a Deadline Slave instance now have a built in concurrent task limit? ie: what stops someone using concurrent tasks instead of slave instancing?

  10. Feature Request. On the right-click launcher menu system, I’d like to start building my own pipeline specific tools and it would be nice to have the option to either create a directory in the repository and then see that name appear as a menu entry, where I cna then add directories containg my pipeline scripts/tools OR alternatively, the ability to have sub-menus under the “scripts” menu entry would be another option. At the moment, I only have the ability to name my py scripts under the “scripts” section in such a way to control their order of display. However, this isn’t that helpful moving forwards and |I don’t what to have to re-factor the names later on due to 1 change! SO, the ability to have a custom menu entry / scripts menu would be nice.

Mike

Thanks for the feedback!

  1. I can’t seem to reproduce this. I right-click on one slave, select the Ping option, and only that slave is pinged.

  2. We were able to reproduce this once, but then never again. We’re thinking it could be that the button isn’t properly refreshed after setting its state to disabled. We think we have fixed it, as we’re now forcing the UI to update. We haven’t been able to reproduce it since.

  3. Actually, non of our sub-menus have icons. :wink:

  4. They all are, and we can definitely change the text to make it a little more obvious. We’re still not sure if we want to allow each slave to have a different “launch at startup” option. The main reason we haven’t done it yet is that it won’t work with auto-configuration because that’s based on IP address (and of course every slave on the same machine will have the same IP address).

  5. Thanks for catching this. In the next release, we will enforce that only alphanumerics, underscores, and hyphens are used. Now that the slave name is no longer tied to the machine name, having an independent slave naming convention does sound interesting. Could be as simple as appending some extra data to the machine name, like you mentioned.

  6. Hmm. We wanted it so that if there are no individual slaves stored on the machine, that it would default to a slave with the original computer name. But I can see how this is misleading. We’ll just stop adding this “default” slave to the list if there are none.

  7. Self healing and redundancy is on the roadmap, just not for 5.1. :slight_smile:

  8. Logged as a bug.

  9. The concurrent task limit per slave is still 16. The multiple slave feature is meant to solve two main situations:
    i) When rendering 16 concurrent tasks of a job still doesn’t fully utilize the machine.
    ii) When wanting to render different jobs on the same machine.

  10. I really like the idea of using subfolders to make submenus. I’ve added it to the wish list.

Cheers,

  • Ryan
  1. OK, to expand on my previous note on this issue. The status does change in the monitor to “Online” under the ping data field column, when you click on the right-click menu entry. However, you should find that if you hit refresh a couple of times, it will go back to saying “disabled”. Pulse was NOT running during these tests. Selecting ALL slaves for that particular machine and repeating the process and it works fine. Can you reproduce it now?

All other responses are cool/I agree with/want v5.2 now :slight_smile:

Thanks,
Mike

Ah, thanks for clarifying. Do you have the “Ping Slaves” toggle enabled in the toolbar? If not, then this is actually normal operation. The right-click ping is meant as a “one time only” type of thing so you can get instant feedback on a particular machine.

Yep, that’s it. Guess this just needs to be well documented :slight_smile:

Privacy | Site terms | Cookie preferences