DR Manager

[Jpjapers]

I cant be the only one constantly wrestling with DR and having to manually start and stop the service or program.
It would be fantastic if corona had a dr manager like backburner where, from a master pc,  i could see what the status of each node is (see if its parsing or waiting or whatever)remotely stop, start and restart the dr service (when running as a service) as well as run the actual drserver executable when necessary.

Im sure theres also alot of other interesting things you could do with it but having a couple of buttons to manage DR and its many quirks would certainly be useful.
If not, is there a way to see the various stats DRserver shows on its UI via the corona API So i can perhaps try and code something simple myself?
Ive been trying where possible to hack together powershell stuff to do this at the moment.

[Frood]

I cant be the only one constantly wrestling with DR and having to manually start and stop the service or program.

Why do you need to do so?

It would be fantastic if corona had a dr manager like backburner where, from a master pc, i could see what the status of each node is

BB does it's job even though being old and retarded, but nothing you mention is actually possible with BB apart from frame progress and job assignment (which makes no sense when talking about DR, or is it exactly that?). Information about nodes provided in the DR tab of cVFB is not enough for you? (Presumed that you are running only one max instance, there is still that bug where only the first running max instance can display them). BTW did you know that you can do a DR render, disable DR in UI, change assigned slaves and switch DR on again while your render is running?

(when running as a service) as well as run the actual drserver executable when necessary.

As fas as I know you are running drserver as service anyway. So you can query/start/stop DrServers easily remotely with

Code: [Select]

sc \\<remotecomputer> query <servicename>
sc \\<remotecomputer> stop <servicename>
sc \\<remotecomputer> start <servicename>

You also may have a look here https://forum.corona-renderer.com/index.php?topic=7890.msg119279#msg119279 for automating stuff for multiple nodes.

Also have a look at deadline which spawns DR servers if necessary automatically and is able to load balance/manage DR servers for a larger group of nodes and users (see https://docs.thinkboxsoftware.com/products/deadline/10.0/1_User%20Manual/manual/app-corona-dr.html, personally I have no experience with deadline).

But I still don't get what the problem is and how you would like to solve it :) Maybe you could describe a case in which such a manager could help? Is it about having a bunch of slaves which should get assigned to masters in a clever manner? Being able to reserve slaves for certain masters for example? Setting priorities?


Good Luck

[Jpjapers]

Why do you need to do so?

Ive been getting "cannot send scene" on some nodes. Some nodes that are just constantly parsing. Some that never start. Issues with Hair and fur (which doesnt show a dialog when running in service mode but does when running the drserver exe, It wont render if you dont click the warning) And it would be great to be able to see the logs to debug janky nodes without having to remote into each one and check the log separately.

BB does it's job even though being old and retarded, but nothing you mention is actually possible with BB apart from frame progress and job assignment (which makes no sense when talking about DR, or is it exactly that?). Information about nodes provided in the DR tab of cVFB is not enough for you? (Presumed that you are running only one max instance, there is still that bug where only the first running max instance can display them). BTW did you know that you can do a DR render, disable DR in UI, change assigned slaves and switch DR on again while your render is running?

The readouts in the vfb have their purpose but if you have a problematic node it would be nice to just be able to restart DR from the manager side rather than having to manually remote in as well as see the stage its currently working on. The VFB readouts dont seem to update as much as the ones on the DR Server UI.

As fas as I know you are running drserver as service anyway. So you can query/start/stop DrServers easily remotely with

Code: [Select]

sc \\<remotecomputer> query <servicename>
sc \\<remotecomputer> stop <servicename>
sc \\<remotecomputer> start <servicename>

You also may have a look here https://forum.corona-renderer.com/index.php?topic=7890.msg119279#msg119279 for automating stuff for multiple nodes.

Also have a look at deadline which spawns DR servers if necessary automatically and is able to load balance/manage DR servers for a larger group of nodes and users (see https://docs.thinkboxsoftware.com/products/deadline/10.0/1_User%20Manual/manual/app-corona-dr.html, personally I have no experience with deadline).

But I still don't get what the problem is and how you would like to solve it :) Maybe you could describe a case in which such a manager could help? Is it about having a bunch of slaves which should assigned to masters in a clever manner? Being able to reserve slaves for certain masters for example? Setting priorities?


Good Luck

Im running drserver as a service yes. But as i say there are some instances where id like to know what my node is doing. Is it stuck on parsing. Can i remotely view the log for that node to debug whats going on. Can i restart the node. Is there an error message that should pop up but isnt because its running as a service (again hair and fur) . Simple tasks. Like i say ive been starting and stopping the service with powershell commands but it would be nice to have a way of doing it within the VFB.

[Frood]

Ok, that warning which has to be clicked away is to be considered a bug (ignoring silent/quiet/network mode) of the hair and fur system it seems (what triggers this?).

As for looking at logs I can only recommend (for various reasons) to use a graphical tail, like SnakeTail. You define logfile views + keyword highlighting once for nodes or a group of nodes ("sessions") and you are able to live monitor the logs directly via UNC or share without that remote login stuff by just opening a single xml file, see screenshot (where that new annoying v4 bug from a launched DrServers first job is clearly visible.)


Good Luck

[Jpjapers]

Ok, that warning which has to be clicked away is to be considered a bug (ignoring silent/quiet/network mode) of the hair and fur system it seems (what triggers this?).

As for looking at logs I can only recommend (for various reasons) to use a graphical tail, like SnakeTail. You define logfile views + keyword highlighting once for nodes or a group of nodes ("sessions") and you are able to live monitor the logs directly via UNC or share without that remote login stuff by just opening a single xml file, see screenshot (where that new annoying v4 bug from a launched DrServers first job is clearly visible.)


Good Luck

Yeah its a bug that was repeatedly reported to AD over the last decade and yet theyve done nothing about it. But running dr as a service suppresses it but doesnt resume the process. You have to stop the service. Run DR Manually and wait for the message. Again not a corona issue.

Ill take a look at Snaketail but all in all i feel there still should be some manager control of your nodes running dr or at least further availability of the information displayed on the DR UI on the manager PC.