Skip to content

Commit

Permalink
Added more fixes
Browse files Browse the repository at this point in the history
  • Loading branch information
balajialg committed Oct 13, 2023
1 parent fa86d66 commit d9c66d1
Showing 1 changed file with 5 additions and 5 deletions.
10 changes: 5 additions & 5 deletions workflow/performance_issue.md
Original file line number Diff line number Diff line change
Expand Up @@ -2,10 +2,10 @@

Certain large courses or courses with complex software/compute requirements can consume lot of memory/cpu which can result in poor user experience for students and/or increased cloud costs. Some of the commonly reported performance issues are due to one or many of the following reasons,

- Students are printing large data frames to a notebook directly or trying show a table that is too large in the notebook. The way to solve this issue is to recommend students not to print large data directly to notebook. You can slice the dataframe to a smaller sample that you want the students to explore further. Please always review your datasets and remove data that are not core to achieving the required learning outcomes for students.
- Students are running a Python/R/Julia code with an infinite loop. You can inform students to check whether any one of their code is running an infinite loop. You can ask students to reach out to you/your team if they have problem with the code. You can also check with the infra team to identify the problematic user notebooks and can access them to identify and troubleshoot the exact issue.
- Students are printing large data frames to a notebook directly or trying show a table that is too large in a notebook cell. The way to solve this issue is to recommend students to not print large data directly in their notebook. You can slice the dataframe to a smaller sample that you can share with your students to explore further. Please always review your datasets and remove parts of data that are not core to achieving the required learning outcomes for student
- Students are running a Python/R/Julia code containing an infinite loop. You can ask students to review their code constantly to check if they are running an infinite loop. You can ask students to reach out to you/your team if they have problem with the code. You can also check with the infra team to identify the problematic user notebooks and do the necessary code review to solve this issue.
- Students are joining tables that are large. Once again, try to break down the dataset to subset that is of interest to achieve the course objectives.
- Students are having multiple notebooks open at the same time across one or many browsers. If students report errors such as 503, 401 error codes etc.. ask them to check if they have notebooks open in multiple tabs. If it is the case, ask them to close the other tabs and just have a single active tab. As a best practice, please ask students to have a single active tab with notebook.
- You upgraded to the latest version of a package without testing it extensively (e.g.: Otter grader). As a rule of thumb, upgrade packages in staging environment and test the notebooks extensively. Only when you feel comfortable with the updated environment, ask the infra team to upgrade to the latest package version in the stable environment. If you are unsure of the URL for your staging environment, ask the infra team.
- You are using databases like SQLite as part of your workflow without having a consultation about the best practices with the infrastructure team.
- Students are having multiple notebooks open at the same time across one or many browsers. If students report errors such as 503, 401 error codes etc.. ask them to check if they have notebooks open in multiple tabs. As a best practice, please ask students to have a single active tab with a notebook and close other tabs.
- You upgraded to the latest version of a package without testing it extensively (e.g.: Otter grader). As a rule of thumb, upgrade packages in staging environment and test the notebooks extensively. Only when you feel comfortable with the updated environment, ask the infra team to upgrade to the latest package version in the stable environment. If you are unsure of the URL for staging environment for the hub you use for teaching purposes then ask the infra team.
- You are using databases like SQLite as part of your workflow without consulting the about best practices with the infrastructure team.
- You are trying to use GUI-based applications like pyqt5 and QGIS without consulting the infrastructure team

0 comments on commit d9c66d1

Please sign in to comment.