Recent searches
No recent searches
Feedback on the New API Usage Dashboard
Posted Apr 11, 2024
Hello Zendesk and Product Manager Sean Newton,
I wanted to share some thoughts on the new API Usage Dashboard you rolled out recently: we were excited to dive into more detailed API usage reports, but there are a few things that could be better.
First off, I noticed we can't attach images in the feedback section, so I'll have to describe everything in words.
Issue summary
We ran into a snag with the Zendesk Support API api/v2/users endpoint on April 10, 2024, from 07:54 to 10:15 UTC. We kept getting 500 errors for about 2.3 hours, until endpoint started working again.
Dashboard syncing
When we checked the dashboard for info, it didn't show any data for that day. Fast forward to 30 hours later, and it still showed zero requests for the previous day. A tiny tooltip mentioned it was last updated way before the outage, which explains why it missed our issue, altough according to the tool tip, it should be an hourly sync. It was disappointing, especially since it was our first real need to use the dashboard.
Missing important operational data
The dashboard's main graph is missing some key info. It shows successful requests and 429 errors but leaves out the failed ones. Including all error types, especially 4xx and 500 errors, would give us a much clearer picture of what's happening.
Unclear or incomplete data
The sections under the main graph don't provide clear actionable data.
Under the timebased graph, we have 2 sections:
- Top 5 endpoints by volume
- Top 5 endpoints by 429 errors
For instance, it says we made 508 requests to /api/v2/macros, with only 161 successful, but the error rate is somehow only 0.2%, with just one failed 4xx request and no 429 fails.
Top 5 endpoints by volume | Total calls | Successful calls | Error rate % | 429 errors | 4xx errors
/api/v2/macros 508 161 0.2% 0 1
Identify the integrations making requests
Additionally, we can't tell which of our integrations are causing the issues since there is no way to filter the presented data by integration or token.
Overall Usefulness
We expected an API usage dashboard would aim to help us avoid failures and improve support efficiency, but it seems more focused on monitoring high API usage than actual operational stability.
It's your call on what to prioritize, but it would be great to see improvements that help us monitor for failures too.
I hope this feedback is helpful, and I look forward to seeing how the dashboard evolves.
Thanks for considering our input.
3
10 comments
Joel Hellman
Tooltip on timeline - please include hours
It would be great if the timeline could include the time period. It was hard to correlate requests to hours using the graph.
Apparently I can insert images in comments.
As you can see, I can see a date but no time information:
If I instead choose “Custom Range” and select one day, I get the state below, and the only option on granularity here is the “Daily”, and I cannot select “Hourly”. I did test clicking on the graph, but it doesn't support drill down.
Since sync is not working, I could not leverage “Yesterday”, which seems to support “Hours”:
But even if that would have worked, we wouldn't always be looking at yesterday. For example, usage during a weekend, or when going back to review recent usage.
1
Joel Hellman
Nice if you can support the Time format from Localization
We are in Europe with 24 hour clock, but dashboard has everything in AM/PM.
Would be great if you support the Time format we have set in Admin > Account > Localization, but it's not the most important feature either.
1
Shawna James
Thank you for taking the time to provide us with your feedback. This has been logged for our PM team to review. For others who may be interested in this feature request, please add your support by upvoting this post and/or adding your use case to the comments below. Thank you again!
1
Joel Hellman
Hi Shawna
Thanks for helping me report it. I can see you created a ticket with me as the requester, and currentt status is “Awaiting your reply” - just want to check in you are waiting for my additional feedback on anything?
Best,
Joel
1
Jacob the Moderator
I came here to echo many of the points Joel already made. Thanks for getting the ball rolling Joel Hellman !
While I like this new dashboard and the insights it is giving, there are some initial snags like periodic gaps and consistent gaps (Yesterday is consistently blank for us) in data, incomplete metrics for unsuccessful requests, etc.
I'm hoping to have more than the top-5 caller/endpoints available eventually - like all callers/endpoints, as well as more drill-in options.
Top-5 callers is a great start, but knowing more specifically the context of the calls - either as drill-in, or a separate list would be awesome! Something that would answer questions like:
0
Jacob the Moderator
I just noticed that the date filter will show results for a date 1 higher than what is selected - this may be why Yesterday is usually showing blank results.
It is the same if a range is selected. E.g. Select 14-18 in the calendar picker, and results will show for 15-19.
0
Joel Hellman
All screenshots here are from ca 2024-04-18 18:30+02:00
The last updated timestamp as I look at it right now at 2024-04-18 18:24+02:00 says data last updated at at 2024-04-17 01:59+02:00. So last updated yesterday, 2 hours after midnight.
Yet yesterday (2024-04-17) shows no data.
Switching to custom range and including 2024-04-17, and I get data, for 2024-04-17.
At the same time it shows a regular amount of data, comparable to a full days worth, not 2 hours worth of data after midnight.
0
Joel Hellman
Showing last 7 days I can show hourly, and it shows this.
0
Joel Hellman
The blank period should be 500 rest errors from Zendesk) but graph only shows 429 and succesfull requests afaik, it looks a bit weird there.
0
Joel Hellman
I agree on the top 5 - I really want to see all, or more (20). Or a separate view of the endpoints where requests have fails.
The Top 5 endpoints by 429 errors, shows 5 requests with 0 number of 429 errors.
0