Read the FAQs
At a Glance
Check here for answers to the most frequently asked questions (FAQs) about HISE, including certificates of reproducibility (CertPro) and Data Apps. The following questions are arranged alphabetically by topic.
IDEs/Jupyter notebooks
Q: Can I upload files to my IDE?
Files can be uploaded from within the Jupyter UI, but we discourage downloading. To initiate an upload, simply click the up arrow found within the top left of the UI.
Q: Can I transfer files to another IDE?
While file transfer between instances is not supported, the HISE SDK has methods that let you upload files to your own private scratch space folder, which you can then download from within another instance. See the file called "QUICKSTART" in the examples folder of the IDE for details.
Q: How can I expand the amount of disk available to my IDE?
While we are in the process of adding this as a feature, this must be done manually by HISE staff. Please contact support if you require more space.
Q: I interrupted a package installation, and I can no longer install it. What’s going on?
Sometimes install.packages can fail so badly that the lock directory is not removed: this inhibits any further installs to the library directory (or for --pkglock, of the package) until the lock directory is removed manually. p_unlock deletes the directory ‘00LOCK’ that is left behind.
Q: My kernel seems to hang or crash. What’s going on here?
First, verify via HTop/Top in the terminal to see if the process is still running or using resources. Low CPU utilization is often a telltale sign of a hung process. Once you’ve confirmed, consider the following:
- Sometimes, the kernel is automatically killed/restarted due to OOM (out of memory) conditions.
- For more resource-intensive operations, consider writing out a script that can be executed via terminal, and save the output to be imported later into your notebook
- Sometimes, killing and restarting an unnaturally long running cell can result in the same cell finishing quickly
- If the kernel goes from 'busy' to 'reconnecting', there may be a dump file in your working directory that explains why.
- If all else fails, a session can be terminated via the UI in the sessions menu, 2nd tab at the far left. Click the ‘SHUTDOWN' button when needed.
Find more info about the kernel states here: https://jupyterlab.readthedocs.io/en/latest/api/modules/services.kernel.html
Q: I’m trying to access the UI, but I get a 524. What’s happening?
This can happen for a number of reasons:
- Instance memory utilization above 85% can cause the UI to hang. Your code will continue processing during this time.
- Instances take time to boot, roughly 5-10 minutes
- If resuming a paused instance, large amounts of memory require more time to thaw, and may exceed 10 minutes of startup time.
Q: I saved some data outside of my home directory, and it’s no longer there. What gives?
Data inside /home/jupyter persists between reboots. Anything outside of this directory gets wiped.
Q: I’m using IE/Firefox, and I’m seeing some weird issues within HISE.
The user experience for the HISE UI/IDE is designed for Chrome. If you’re using Chrome, then a refresh or relog often resolves the issue. This is often due to fresh updates being pushed to HISE.
Q: My notebook/cell is unresponsive, the kernel seems to be inactive, and I still have access to the terminal. How can I get this back to a working order?
We've found that, from time to time, a running cell can go rogue. This seems to happen more often when the cell is expected to error out. The immediate work-around is to open a terminal window, and kill the process:
- Run 'htop' to find the running process. If you have no other cells running, you can usually identify it by looking for CPU or memory utilization
- Find the PID for this process. This usually shows up on the left side of htop.
- 'Kill' the running process. We'll use the PID we found, and run a 'kill <PID>'. If the kill does not immediately remove the running process, you can force this by running 'kill -9 <PID>'. NOTE: avoid running a 'kill -9' if you're writing to a file, as this may corrupt data.
This appears to be a transient issue, and is often remediated by simply retrying the cell, though killing the runaway kernel does delete objects in memory. To prevent this in the future, we’d suggest the following:
- Checkpoint your notebook by running saveRDS or a similar function to save objects in memory
- For more troublesome code, try running via the R shell in a terminal session
Requests
Q: Where do I find a list of proposed and current user memberships?
For current users: Manage Users Dashboard
For proposed users: HISE Rosters
Q: Can I, the user, have access to a specific bucket or watchfolder?
Assuming the requesting user is already a HISE user they can be granted access to the Watchfolder for their Partner. Requests for non-watchfolder bucket access should be forwarded to the DevOps team for access management. In order to grant access to a Watchfolder, follow these steps:
Open the Accounts page.
Choose the account to use (usually PrimeCollective).
Go to the Watchfolders tab and find the corresponding partner watchfolder.
You can usually determine the user’s partner based on the email domain, but if you are unsure, go to the user list to see what memberships the user has.
Click Edit and add the user email to the list of authorized users. (If the user’s email doesn’t appear in the list of available authorized users, make sure you're in the right watchfolder for the user’s partner, and check the user account as well. If still can't locate the user, message stark.pister@alleninstitute.org.