"Failed to fetch" errors on all flows

[billblais]

I have a number of flows. All have been running fine until today. I am now getting 'Failed to fetch' errors for all of them. All my connections show as invalid in a given flow, but look fine in the main Connections view. I have tried creating new flows, re-establishing connections, using new connections -- all to no avail. I've looked in the posts here and found a few references to this, but no actual answers or solutions other than 'it started working again' or no resolution at all.

Is this happening to anyone else?

[v-yuazh-msft]

Hi @ billblais,

Are you still having this issue now?

There is post with the same issue as you:

https://powerusers.microsoft.com/t5/General-Flow-Discussion/Failed-to-Fetch-GetTable-error/m-p/63762...

As @ aporochine01 said, this maybe a temporary error, please try again to see if the issue still exists 

Regards,
Alice Zhang

[billblais]

Alice,

Yes, it was a temporary problem. It took about 6 hours to go away, however, and for no clear reason. On the up-side, my time- and action-specific flows during that time frame were apparently still 'aware' and therefore kicked off/caught up when the issue went away, but that was well into the evening, causing an obviously unexpected experience for those working with those flows.

What is the cause of this issue? 'Wait and try again later to see if it goes away' is definitely not a solution. Given that there are at least a few reports of this issue, there should be at least something in the way of actual explanation of what's causing this and how it is being addressed. Flow is an impressively powerful tool that we want to build business workflows around, but if it can't provide dependable reliability, that critically undermines the tools usefulness.

To be clear, I am not trying to shoot the messenger, but this is a major uncertainty.

[Anonymous]

I'm having the exact same problem with "Failed to fetch" errors on all of my flows. They were working fine earlier this morning and suddenly stopped working. Is there any solution to this problem besides just waiting several hours? 

[PabloM]

I'm having this exact same problem with my flow. The thing is in my case the problem does not go away.

I need this solved to put in production a really long flow. Maybe someone else had the same problem?

My flow has over 30 excel actions and all connections get broken they also throw an 'invalid connection' error which I haven't been able to solve. I have to reset all connections and then they seem to work for like 20 minutes or so to get broken again. It's an ongoing problem and it won't go away.

If that doesn't happen then I get a failed to fetch error in some actions while the rest still seem to be working properly. It's really annoying and it's simply plain impossible to deploy my solution to production like this. I need a solution, thanks.

[PabloM]

Hello all,

I've been thinking of this problem at large since I had it too. I've run some tests with Fiddler which lets you monitor all http requests done (among other things) when opening web pages or software since I thought that maybe I the problem could be in too many calls being done by the connectors to the microsoft APIs and being throttled as a result. This turned out to be true since when I opened my Flow in edit mode while monitoring the calls made to APIs I could see that some of them were returning the 429 error which corresponds to too many requests.

This means that microsoft is stopping me from continuing doing calls to them and getting all the info I need for my flow to load all the pertinent data and show it to me. That's why I was getting the Failed to fetch error.

I had previously whitlisted all the domains and IP addresses listed here as it was indicated in microsoft documentation to rule out this being the problem: https://support.microsoft.com/en-us/help/4557620/client-request-aborted-or-failed-to-fetch-error-in-...  

Since AFTER doing that the problem persisted I thought on how many calls where my connections making to the microsoft APIs and if this could be the culprit and it effectively was.What you want is not to have a single connection per service used but many to distribute the calls and not get throttled.

There's 2 solutions to get out of this problem, the first one is to create more connections for the same service within the flow you are writing. So you would have to go into the elypsis of one of your actions (of the problematic connector, in my case Excel ones) and from there create extra connections. Once you've done that you can randomize the use of those between all the actions that are from the same service (ie excel). 

That will mean that your actions will use different connections to make the call and will not get throttled.

The second solution is to partition your original flow. This would mean build a squeleton flow which would then call different parts of your original flow. This approach has to be handled with caution since before starting you need to make sure to know how to handle data through JSON and also since understand if you (and your users) are all licensed up properly (in case you are thinking on using HTTP requests to move data between the different flows since that is a premium connector).

The way to do this without a Power Automate per user license would be to use the Solutions section of your Power Automate studio. 

A solution is  basically a container that allows you to save different related flows into it. The advantage of doing this is you can then call all flows contained in your solution from one another. You'd do this using the CHILD FLOW action which will only appear if you have saved all your related flows inside the same solution (AND created them inside the solution as well and not outside and then imported them in, since this can bring problems as well).

Hope all this info is helpful since it took me days to solve all this.

Regards,

Pablo

[AhmedMaher]

If you have Firewall
May you need to exempt this URL "msmanaged-na.azure-apim.net"