nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

June 2026 Upgrade Effecting FME Processes

YRP_GIS

Has anyone noticed that since yesterday morning, any data that is being processed via FME to ArcGIS Online is not reading the full dataset?

We have a dataset with approximately 244,000 records in it. FME can only read from the ArcGIS Online reader 100,000 of those records. When we set the minimum to 100,0001, FME states that it cannot access anymore features. We have also noticed, that as a result of this, our numbers in our source Hosted Feature Class in AGOL has duplicated and triplicated the number of records that we are supposed to have.

image (3).png

EdiePunt_1-1745172616925.png

Encontre mais publicações marcadas com

Content Management

Data and services

Accepted answers

EvanPerryTranspower

Not sure what ESRI Support suggested but what work for me was to go to the items overview page in AGOL and click the link to the REST Services Directory. Then hit the Admin link top right of the page. On the next page pick the Update Definition link at the bottom of the page. In the Update Service Definition dialog box scroll down until you find the maxIdsCount and change it to a number greater than the number of records in the feature layer. Hit the Update Service Definition button.

All comments

abdhassa24

Yes I noticed this as well. It seems like they added a new property called the maxIdsCount and is set to a 100k limit for a single request when retrieving the ObjectIds of a feature service. Before the June update, you can query all the ObjectIds in a single request. Until Safe Software provides a fix in FME, you have to use the Query API with pagination to retrieve all the records in the feature service.

Scott_Sambell

Yes, we have woken up to several clients asking why their feature layers have gone crazy over the weekend. It's affecting hundreds of users here so far. Trying to figure out now the amount of clients/systems/feature layers/ users which are affected by this

AnneShanahan2

Hey - we're seeing this too. As a work around at the moment we need to use multiple feature readers with ObjectID where clauses so we can read in 100,000 feature increments. Luckily our datasets have are not over 200,000 features so is doable but imagine not a viable work around for other situations. And is also easy to go unnoticed on FME Flow as the process still runs successfully so easy to miss that not all feature were processed. Be good to see this fixed asap.

EvanPerryTranspower

Unfortunately yes hitting the exact same problems

sosero

Hello @YRP_GIS , @Scott_Sambell , @abdhassa24 , @AnneShanahan2 @EvanPerryTranspower ,

This was logged as a bug Internally. Feel free to open a new case with technical support, and the bug can be attached to your cases for your reference. Thank you.

Katie_Clark

@sosero Is there a bug number associated with this so users can follow the status of the case?

Clint_Lusk

I don't have anything useful to add, other than wanting to add my voice to the issue.

The dashboard and FME changes alone have stopped my customers from being able to do their normal reporting with our tools. This might push them to use Tableau or some other solution if we can't get this reversed in time.

Clint_Lusk

Correction...

Our data team just found that the large datasets writing to AGOL from FME with over 100,000 records were stopped right at 100,000, but the large datasets that were built using change detectors to only right the most recent data were still piling in there as per normal.

So the solution we're looking at is to add change detectors to all our FMEs writing to AGOL so they're only adding smaller amounts of data.

Still won't help with Reading from AGOL... but it's something.

YRP_GIS

Yes, this is exactly what happened on our end. We submitted a ticket with ESRI, and they were able to assist us with reverting the max record count. Once we changed the value of the maxRecordCount in the REST API, we were able to run our scripts as expected.

EvanPerryTranspower

anthonynorrell

Hi,
Would you be able to share the steps to update this?
I just tried going through the Update Definition section for the affected sub-layer, but the save didn't stick. I tried increasing maxIdsCount from 100000 to 400000 and setting lastEditDate to null.

DeanPetricevich

It has to be done on the Update definition on the service itself not the sublayer, then it applies to the sublayers. When doing on the service there is no need to update lastEditDate.

abdhassa24

Delete By Author

abdhassa24

bharold

Hi Evan, I think you'll find that you can only bump up maxIdsCount to a maximum of 400000, at least that is my experience.

maxIdsCount may now be 1 million.

SiennaEmery

Hi all, just wanted to share that Safe Software has published a Known Issues article tracking this problem for those using FME. You can follow it here for updates:

https://support.safe.com/hc/en-us/articles/47037080794381-Known-Issue-ArcGIS-Feature-Service-Reader-Limited-to-100-000-Records-After-ArcGIS-Online-June-2026-Update

bharold

Thanks Sienna, and also please see this custom transformer implementing Option 3. This was authored for ArcGIS Pro Data Interoperability 3.7 and FME Form 2026.1. If you are using earlier versions please comment and we can provide advice on handling issues, or rework the transformer.

EvanPerryTranspower

Interestingly we have a workbench running daily in Flow that has been completely oblivious to any of this. It’s reading over 900k records from publicly available AGOL feature layer with no issues whatsoever.

It uses the old “Esri ArcGIS Server Feature Service (replaced by ARCGISFEATURESERVICE)” maybe it works differently from the ArcGIS Online readers.

However, testing this on some other layers they are hitting the maximum number of records, mysterious ...

KhaledHassen

FYI, we are increasing the limit to 1M ids. Pl. let us know if you see any timeout in your query that returns large number of ids. Changes should be on in prod tonight if all are good after testing. You can check the layer metadata and that should return you the 1M limit in returned Ids.

Eddie_Idolyantes

Pl. note we have increased the max allowed objectId returned in online to 1 million. The change will be on production tonight.

RenatoSalvaleon3

Here's the Bug link @Katie_Clark : https://support.esri.com/en-us/bug/the-default-maxidscount-prevents-users-from-querying-th-bug-000185853

The article is updated to the latest fix allowing maxID = 1,000,000 records.

HelenWhiteley_tfl

Hi Bruce. I have just hit this issue and am looking at implementing your FME workarounds. However, the feature service I am reading does not belong to me, is secure and has millions of records. I only want a subset of the millions (around 4 million) and had previously achieved this by filtering and saving a subset and then reading using the (new) ArcGIS Feature Service Reader and reading it this way. Is there a way to incorporate the necessary token and query into the http caller method?

bharold

Hi Helen, yes sure. Add the ArcGISOnlineFeatureGetter to a workspace and populate the service URL and web connection parameters with your details.

Then right click on the transformer and choose to "Edit Embedded Transformer". You'll get a new tab in Workbench. In that tab in the bookmark labelled "Get Count" there is an HTTPCaller transformer, edit its where query string parameter to select the data you want from your big service. Make the same where query string edit to the HTTPCaller in the "Read Service Layer" bookmark, then save the edits, and optionally close the tab opened for the edit. If you want, make the where query string a parameter so it's flexible going forward.

VCGIvanBrown

Identifying and re-wiring all production Data Interoperability workspaces that are impacted by this change requires a lot of expense (labor). Some workarounds involve replacing one reader with multiple transformers--adding complexity to workspaces--complexity that would otherwise not be necessary.

RenatoSalvaleon3

We understand your concern and we still have an open item in our back log to keep track of support issues that continue to report this behavior that require more than a million features. Have you tried the single custom transformer solution ArcGISOnlineFeatureGetter_2 yet? You'd probably have to hydrate the schema and I understand that's the harder part.

Unfortunately, even this workaround provided through FME Hub with 23 downloads as of today, may not be sufficient to justify the balance between performance and product functionality.

If you have not migrated to the ESRI ArcGIS Connector package for your current workspaces that use the legacy ESRI feature service format reader and writer yet, this would be your best opportunity to do both. The legacy formats will be hidden in Pro 3.8/FME Engine 2026.3, thus, a migration would be timely.

Participe da Conversa

Membros conectados podem postar, seguir atualizações e mais. Novo aqui? Registre uma conta gratuita.

Links Rápidos

Community Help
Find useful guides, FAQs, and documents to help you navigate and make the most of Esri Community.