AWS Thinkbox Discussion Forums

Upgrade to Beta15 (50661)

I remember last time I upgraded the Repository I had problems (upgrade wont work, installed into wrong place) and in the end deleted and re-installed from scratch.

So, Installation Directory… I assume this is the installation directory is for the Repository. If this is correct, can we change the title to Repository Installation Directory?

So point this to my existing repository directory (/mnt/DeadlineRepository6), then select connect to existing MongoDB, then on the MongoDB settings assume these are correct as installed with defaults and the installation fails with

"
An error occurred while configuring the database

An error occurred while trying to connect to the Database (licserver:27017).
Its possible that the MOngo Database server is incorrectly configured, currently offline, or experiencing network issues.

"

The DB is running as Deadline is currently running…

Suggestions please…

Chris

Yes, this is the repository installation directory. I’m wondering though if making it explicit really necessary, since you are running the “Repository” installer? Was it just confusion over which path to choose when performing an upgrade? If so, maybe we need to add some help text to that installer page to make that more clear.

Which Deadline applications are currently running? Monitor, slave, launcher, etc? That’s really strange that they would be able to connect and not the installer, since they are all actually using the same code to connect under the hood (the installer runs deadlinecommand when configuring the database).

Which machine are you running the repository installer on? Is it one of the machines that the Deadline applications are running on? If not, maybe try one of those machines to see if it makes a difference.

Yes, this is the repository installation directory. I’m wondering though if making it explicit really necessary, since you are running the “Repository” installer? Was it just confusion over which path to choose when performing an upgrade? If so, maybe we need to add some help text to that installer page to make that more clear.

Yes…I know I am running the repository installer however the confusion on my part is that am I installing the Repository at that point or the Deadline software as they go into to different places.
Suggestion would be to change the screen text to Repository Installation Directory.

The repository is running on our license server (ldap server, DNS server, etc etc…). I would like this to be the Pulse server also but I am not there yet.
As far as I can see there are no deadline processes running on that system.
I have 20 slaves (render nodes) attached to the DB, and maybe one or to monitors also connected from other workstations.
Starting up the deadline launcher on the license server system gives me cannot connect to the repository/MongoDB errors although its pointing to the correct location (which is local to itself), and deadline launcher/monitor work if fired up from another workstation.
Checked mongoDB as thats running as expected.
bringing up licserver:28017 gives me a webpage for MongoBD…plus log details for yesterday and today (so far).

I would really like to resolve the upgrade, although I need to move on and get everything on the same version of software as the slaves are on 50509 while the workstations are reporting 49986 and will not update either as I get “connection refused” errors which then kills ALL of my terminal windows shortly after starting (not nice at all!).

If I cannot get the upgrade to run through then I will trash the current repository and start again…

Where did you install the repository to on the license server? Was it the default /opt/Thinkbox/DeadlineRepository6 location? If so, are you picking this location when you are running the repository installer on the license server? Or are you pointing to /mnt/DeadlineRepository6 (which you mention further up)? If you are pointing to /mnt/DeadlineRepository6 instead of /opt/Thinkbox/DeadlineRepository6, maybe that’s the root of the problem?

Also, as I mentioned earlier, you could try picking one of the workstations or nodes that’s currently connected to the database, and run the repository installer on that machine. Choose the repository path that the node is already using as the installation directory. In theory, if the Deadline applications on that machine can connect to the database, the repository installer should have no problems configuring it.

I really want to use the current beta location of /mnt/DeadlineRepository6 which is where the repository is currently configured (and working) on the license-server (its the same mount point on the linux workstations)
Having to change to /opt/Thinkbox/DeadlineRepository6 means I have to go thorough samba changes and export changes, plus change the mount on 40 systems before moving onto the Windows and Mac systems.

What I dont understand is why the local Deadline software cannot contact the mongoDB while everything else can?

Okay, that’s fine. If /mnt/DeadlineRepository6 is the place you originally installed the repository to on the license server, then there shouldn’t be any problem with this setup. It just seems odd that a Deadline app complains about this repository when running on the license server, but is fine with it when running on a node.

So maybe just try my second suggestion about running the repository installer from a node, and choosing /mnt/DeadlineRepository6 as your install directory.

Also tried the upgrade on the license-server pointing to the default location of /opt/Thinkbox/DeadlineRepository6 and connecting to existing mongoDB - fails with unable to connect to DB.

Then tried installing Deadline Repository on my test workstation using /opt/Thinkbox/DeadlineRepository6 and connecting to existing mongoDB. This connected to DB - cancelled installation.

Tried again using the /mnt/DeadlineRepository6 (which is the current network mounted location) and this also connected to DB, so as it was pointing to the correct location I let it run. Installation run through so I started up deadline launcher which detected an upgrade.
This ran through the upgrade and then fell over with
‘Import site’ failed.
Then the shell died.

Ran a second time from command line and got the same error, then it tried again and completed correctly as reported by help -> about deadline6.

So we have progress, but not being able to run anything on the license server is going to make using the license server as a pulse server impossible.

You’re going to have to run the client installer on all of your nodes/workstations again since they had much older versions on them. Once they’re all on beta 15, auto-upgrades should work going forward.

Also, you will have to run the beta 15 client installer on the license server to run pulse on it. After you have installed the client, try running pulse, and then send us the error you get on startup.

Thanks!

  • Ryan

Render farm upgraded OK via deadlinelauncher.
Workstations - these I have to do before people start in the morning, so its an early start for me on Wednesday.

Thanks for the help in getting this resolved.
Next is Pulse, central config, and QuickTime creation.

Privacy | Site terms | Cookie preferences