Status updates for Canvas Data

[s_moles]

I work on a team that collects and analyzes Canvas data for our university.

The Instructure status page (status.instructure.com) is very helpful, but, so far as we can tell, it does not include the status of the Canvas Data service. It would be great to have the status of Canvas Data, including any known issues, similarly posted and clearly identified as pertaining to that service.

Example of why this would be useful

We run a nightly Python-based job that downloads the files for all tables except the Requests file. Over the past few months, the job has failed several times. We've spent many hours investigating the failures and it now appears to us that they are due to connections closed by AWS and quite possibly not because anything we are doing on our side. Had we been able to check the status of the Canvas Data service, confident that known issues would have been reported, we might have avoided much confusion and lost time and arrived at a solution more quickly. 

[briandgauvin]

Having some indication of whether the Canvas Data service is running smoothly would be extremely helpful.  Our team has spent many hours looking into failures thinking we were doing something wrong, but may have potentially been due to the service not working correctly that morning.

[scottdennis]

Hi Brook,

Thank you for coming to the community to provide this feedback.  We usually reserve the feature conversation process for conversations about improving Canvas itself so I will go ahead and archive this submission.  However I know that we currently have a team that is investigating how best to improve the status and statushistory pages.  I will take your request to their team lead and make sure they understand the request.

Thanks for your time and input.

[pgo586]

I also agree that having this information available would be extremely useful. We've also experienced downtimes that  affected production tools dependent on Canvas Data.  Having a status page - with clear information as to the current issues and expectation for resolution time- would have likely deemed support tickets unnecessary, while it would have also saved us from having spent unnecessary resources towards debugging. 

[jeff_longland]

+1 - we encounter the same problem with gateway time outs and dropped connections. Short of resolving those problems, it only seems fair to acknowledge Canvas Data service interruptions. Clients don't distinguish between "Canvas itself" and "integrated products developed by the same company and delivered under the same product name". That's a weak response. Think a bit more about the user experience. The internal organizational boundaries are not your clients concern.

[lfeng1]

Hi @scottdennis glad to hear there is a team looking at improving the status and statushistory pages, and that this idea is being brought to them.  I think it is still valuable to be able to track this type of request somewhere visible to all customers.  Many schools are dependent on the Canvas Data feed each night and when it is down there is currently no good way to tell.  Appreciate any help you can provide here.  Thanks!!  

[scottdennis]

Hey Linda!

I can see the value in a place for ideas about Instructure things that aren't necessarily Canvas - like when we organized ideas for things partners might want to choose to build.  Let me talk with the team and see what they think.

Thanks,
SD

[lfeng1]

sounds great - thanks Scott!!  Linda

[Jeff_F]

@s_moles  -

While indicators such as https://status.aws.amazon.com/ (RedShift) and https://trust.tableau.com/ are available, it seems that when we had an issue it was due to the file update process failing and these sites do not give us that info.

As we are hosted, Canvas Data support is available but I found it easiest to create a calculated field on discussions - something like MAX (timestamp) to give us an idea of data recency.  If that report isn't running then I note the issue and communicate to set expectations that the service is not available for a day.  Invariably the system corrects itself the next time the data updates, which is currently 24-36 hours.  But I send a note to support regardless in case the issue is something else. In 2019 and 2020 this happened just 2-3 times, which is much fewer occurrences than in 2017-2018.  I suspect the large Requests table was to blame for much of these.

Looking to the future with Canvas Data 2 being set to update the data every 4-6 hours I am thinking this is less of an issue. Even so, it would be nice should the prior set of data be available should the update fail. That way reports will still run, albeit with slightly older data.  

Thanks is reading!