Frequently asked questions¶
How does licensing work?¶
All licensing is taken care of by us. We have metered licensing agreements with several major vendors. You can check the supported software page for the full list.
If your renders rely on 3rd party software that requires a license, we can't connect to your license servers, however, you should submit a support ticket as it may be possible for us to make arrangements with the vendor.
Does Conductor support renderer X or plugin Y?¶
If the package isn't listed in supported software page we don't currently support it. Feel free to reach out to us at support and inquire about adding it to our service. If you wish to use a version that isn't listed, also reach out to support.
I would like to do a test / I'm student / I'm a non-profit. Can I do it for free?¶
Conductor doesn't own the underlying infrastructure. We're built on-top of GCP and AWS and inccur their compute costs whenever jobs run on our service. Therefore we can't offer free tests beyond the introductory credit.
What OS do you run on?¶
All of our jobs run on a Debian 9 (Stretch) image.
Are you integrated with any render farm management systems?¶
Can I run Conductor locally?¶
Conductor doesn't have a local worker or server installation. However, please reach out to support so that we can discuss possible options to extend your render farm to Conductor and leverage the untethered scale and performance of the cloud.
Is Conductor available on the GCP/AWS Marketplace?¶
Yes it is!
Using the service¶
What is the sync failed error?¶
The sync_failed error occurs when the process (syncing) of transferring a job's files to high-speed storage fails. Sync failures are not associated with user error and there's no customer cost assoicated with it. These errors are rare and intermittent. Navigate to the job in the Dashboard, right-click and select "retry sync" to retry the sync process.
Why is my job stuck in pending?¶
Pending indicates that your data has been transferred to the high-speed storage and waiting for machines to become available. Jobs are started on a two-minute cycle. If your job hasn't started after two minutes, it's likely that the selected instances are unavailable in sufficient capacity on the cloud provider. You can try changing instance types or switch from preemptibles/spots to standard/on-demand machines.
Why are my tasks on hold?¶
Tasks can be put on hold for various reasons:
- You submitted your job with scout frames (all Tasks that aren't scout frames will be on hold)
- You manually set your job/task to Hold
- There's a billing issue with your account.
For the first two reasons, you can unhold your jobs manually by selected the job(s)/task(s), right-click and select unhold in the dashboard. For billing issues, please contact support if you require additional help.
Can you priortize my jobs?¶
We don't (and can't) prioritize one customer's jobs over another. The best way to get your jobs running quickly is to choose non-preemptible instances that aren't in high demand. You can prioritize your own jobs amongst themselves by setting the priority field. A higher number indicates higher priority.
Why did my job fail?¶
A job has a failed status if it returns with a return code other than zero. There are many reasons why a job may fail. The most common are:
- The software in the task has an incorrect setting. Double-check all your settings and do a test-run locally. Many software packages can exhibit different behaviour when run interactively or via the command-line.
- Files are missing. Check the render logs for errors related to missing files. See the question My job is failing. The render logs show that a file is missing. What do I do? below for more details.
- The wrong instance type was chosen. For performance reasons, render instances don't have any SWAP space. If the task requires more memory than is available, the task will fail. Therefore, it's important to choose an instance type with sufficient memory. If your job requires a GPU, ensure a GPU-enabled instance is selected.
Can I create job dependencies?¶
No. At this point in time we don't natively support job dependencies.
What instance type should I choose?¶
This is highly dependent on how the selected software packages scale and are licensed. In many cases, the software scales linearly (to a limit) and the license is per-instance (as opposed to being per-core). In these cases, it's best to choose an instance with the highest-core count available so that your job completes as fast as possible (and at less risk of preemption), reducing the the license cost. See our article for more details.
Will I have to re-upload my files for every job?¶
No. We use a smart caching mechanism. If you've uploaded a file and its contents haven't changed, our uploader will recognize that and skip over it. Even if the path or filename has changed, it won't be re-uploaded as our system will reckognize it as the same file.
My downloads and/or uploads are slow. What can I do?¶
Start by using a speed test (ex: www.speedtest.net) to verify that your transfer speeds match what's promised by your Internet Service Provider. If not, try restarting your networking equipment, and make sure that no other device on your network is utilizing too much of your bandwidth at the moment.
When reading specifications and results, keep in mind the difference between Megabits and MegaBytes.
If the results of the speed test is within the expected range, monitor your network throughput while running the uploader/downloader. Every OS ships with a native monitor. If the throughput isn't within range of the speed test, here are a few more things to check:
- Many small files tend to transfer a lot more slowly than a single large file.
- Ensure that the storage you're writing to or reading from can keep up with your internet bandwidth. Pay particular attention to non-local drives such as Dropbox, GoogleDrive, etc...
- Ensure that your router isn't limiting traffic from the Conductor uploader/downloader
- Ensure that your NIC is able to maximize your internet connection (ex: 10Gbps internet connection but only have 1Gbps network adapter). If reading/writing from a network drive, you'll need to account for the bandwidth usage there as well.
How long should it take my job to start?¶
Once all the files have been uploaded, your job will start syncing to our high-speed storage momentarily. This is usually a a quick process but can take a few minutes with a large amount of data. Once the syncing has completed, your job will wait for machines to become available. Once they, the job will start running within a few minutes.
My job failed. Can I get reimbursed?¶
We can't take responsiblity for jobs that are submitted with incorrect settings, etc. It's important for you to verify everything prior to submitting. We encourage the use of scout frames to ensure that everything is working properly on a single (or few) tasks before spinning-up many machines and potentially incurring a large cost on a failing job.
My job is failing. The render logs show that a file is missing. What do I do?¶
- Does the missing path from the render logs exist on your local machine? If you're on Linux or OSX, it should be an exact match. If you're on Windows, it should be an exact match with the drive letter stripped. If this isn't the case, you'll have to investigate why the DCC modified the path.
- In the dashboard, navigate to the task and expand Uploaded Files. Ensure that the missing file is in this list. If it's not, it's possible our dependency scraping skipped over the file. Add the file in the extra uploads section of the submitter and feel free to drop us a line at support so that we can add it to our dependency scrapers.
- If the above two points don't solve the problem, please reach out to us at support and we'll investigate the cause.
I went back to look at some old task logs and nothing shows-up.¶
We only retain task logs for 30-days. After that, they're no longer available.
My task is running but I don't see any logs for it¶
It may take a few seconds up to 5 minutes for the logs to be made available on the dashboard. If you still haven't seen any logs after a couple of minutes of the job running (and you've tried refreshing) feel free to reach out to us support so we can further investigate.
Can I use Arnold/Vray/etc... standalone?¶
Yes! Select one of the renderers plugins as the only package (ex: MtoA) and the standalone renderer
will be available. See this page for an example on how to use
How do I update my payment information?¶
The account owner may change the billing information via the Payment Information tab on the Admin page of the web UI.
How do I add or manage a user?¶
The account owner may add users via the Users tab on the Admin page of the web UI.
How do I tell how much of my cost limit I’ve used?¶
Check the Account Cost Limits tob of the Admin page page of the web UI. You can see your project and account limits, and how much of the overall limit you've been charged to date.
How do cost limits work, and why would I use them?¶
Cost limits allow you to maintain control over the costs incurred on a project or account basis.
How much did Job X cost?¶
Per-job costs can be displayed on the dashboard as the
cost column on the jobs lists. To enable it,
press the Columns button (top-right beside the search bar) and select
Do you offer any discounts?¶
Yes! We offer discounts based on pre-payments. The more you pre-pay, the greater the discount. See the Payment page on the Admin tab in the dashboard to see more details. If you're an enterprise customer, feel free to reach out to our sales team.
How am I charged?¶
We only charge for the time a job is spent running. We don't charge for your jobs while they're in the pending or syncing states. The charge for a job is an aggregation of instance cost, license costs and the Conductor service cost - all charged by the second. In addition you are also charged $0.18/GB/month for storage (pro-rated). See the Pricing page for more details.
Once your project is complete you can request to have all your data purged via the Storage tab in the dashboard (admin rights are required). This will avoid any future monthly charges for storing your data.
Do you have an API?¶
Yes. You can install our Python API (ciocore) from PyPi
pip install ciocore
My API key stopped working. What happened?¶
Only a single API key can exist for each account. If you've generated a key and it's no longer working, it's likely someone else generated a new key - which would invalidate all previous keys. If that wasn't the case, try generating and using a new key.