Convert or Create PDF is failing to create a file but not actually failing in the flow

[dsagfhjj]

I have several flows that take html detail from various powerapps and convert them into a pdf to email out as reports. My flow keeps failing on the email because the attachment is blank when it expected any data. The convert html to pdf and create pdf steps are not reporting as failed, but the pdf file doesn't appear to be created in the selected one drive location. I've confirmed this with several users, flows, and powerapps that perform similar functions to provide reports to users. These flows worked fine as of yesterday.

Is this a system wide failure or was some update pushed that I can't find a notice for?

[JLutchen]

I'm also having this issue, it first appeared between 2:00 and 3:00 Eastern US Time yesterday (July 10th)

[dsagfhjj]

I'm eventually raising an error when the email attempts to include the attachment that doesn't exist. Where are you getting the error?

My users report the last successful run as 3:22pm EST.

[Jarodamus]

I am also getting this error. Seems to be isolated to the "Convert File (Preview)" action.
I am trying to convert some exported emails as .eml files to PDF files.

Re-submitting previous successful jobs yields the same result, empty files.

[JLutchen]

The error is happening in the send email portion of my flow (which is right after the convert to PDF action) but the cause is the convert PDF action not having any file content in the output section.

[dsagfhjj]

I eventually found the same thing. I eventually found that the PDFs are being created, but they're corrupted/empty.

[dsagfhjj · Accepted solution]

It is working today. I changed nothing.

[AlexEncodian]

@dsagfhjj / @JLutchen / @Jarodamus 

Saw a similar thread today...it may be a system / connector issue. I would raise a ticket with Microsoft and hope they can fix it soon.

An alternative in the short run is the Encodian connector (Standard connector but 3rd party...has a free tier).

[adamp1]

It seems Microsoft arbitrarily and silently changes the backend engine that actually does the PDF conversion without notifying anyone. Recently they switched from a very old "PDF Sharp" to a more modern Chromium, which was a welcome change (despite being done out of the blue). We modified all of our HTML that was being converted to support very nice modern CSS, then just today, it seems to be back to PDF Sharp. Both of these events have caused catastrophic interruptions in our production products that we use with PDFs. 

It would be great if Microsoft was more transparent about these changes. I'm sure their argument will be just not to use "Preview" connectors, but "Preview" operations feel like they make up the majority of useful operations, and they've been in Preview for several years.