Forum Discussion
GEN2 dataflow error - We received multiple duplicate continuation tokens from Azure Storage
I spoke with a Mindtree engineer. The release train went to production on 4/5 and this unusual error appeared to us for the first time in our daily GEN2 refresh on 4/7.
They want me to open a new ticket (SR). After that Mindtree promised to send the bug to Microsoft via ICM. They claim there are no prior ICM's with a similar error message yet:
We received multiple duplicate continuation tokens from Azure Storage.
I will probably send this issue thru pro support (Mindtree) once our users start to notice/complain. It may be a few days for them to realize their stuff is broken (again). The worst part is that this new problem came up in the same "train" that deployed a fix for a different problem. A very LARGE part of my job these days involves talking to Mindtree support about Fabric bugs. Some days I find it very troublesome to be a Fabric customer.
Hi dbeavon3 ,
We are following up once again regarding your query. Could you please confirm if the issue has been resolved through the support ticket with Microsoft?
If the issue has been resolved, we kindly request you to share the resolution or key insights here to help others in the community. If we don’t hear back, we’ll go ahead and close this thread.
Should you need further assistance in the future, we encourage you to reach out via the Microsoft Fabric Community Forum and create a new thread. We’ll be happy to help.
Thank you for your understanding and participation.
Hi v-menakakota
Can you tell me what org you work for or why you keep following up? I think it seems likely that this topic is regularly searched, and this page is regularly viewed. True?
You should know that customers invest far too much time opening SR's with Mindtree. A ticket like this one would EASILY take me 2 or 3 weeks of effort with Mindtree, and there is absolutely no pot of gold to be found at the end of that effort. Based on dozens of past experiences it is NOT likely that the PG would quickly fix the bug, nor create a "known issues" page for it. As such, this is the type of investment I am NOT willing to make right now (until there is an urgent outage that forces me to that point.)
Furthermore, I believe the PG could easily monitor their telemetry, and measure the number of times this issue happens. They could find a pattern and fix it, or simply add it to their known issues page (on their own initiative). The PG's could take the initiative on something like this if they wanted, and they certainly don't need my help to fix their bugs.
I have an informal contact in the PG and I think the best guess is that there is some sort of error coming out of the lakehouse, when interacting with parquet files. These implementation details are buried under several layers of abstraction. Again, it would be easy for the PG to find the error message in their source code, and it would be easy for them to build realistic samples that apply stress to that part of the code. If they don't want to do all of this work themselves, they should share the source code openly. Fabric is based on massive amounts of open source code (including all the stuff related to parquet files, so ther is no reason they can't share a little more code and allow the community to find/fix this bug).