we are facing a severe performance issue with Sales invoice posting in our D365 Finance & SCM production environment and would appreciate any hints or similar experiences.
Scenario
- Environment: D365 F&O, PROD, Application release 10.0.45, Platform release PU69 (7.0.7690.69)
- Process: Sales invoice posting batch (Accounts receivable --> Batch Invoicing)
- Until around 13 November 2025, we were able to post about 23,000 invoices in 6–8 hours in this environment.
Symptoms
- After the change in behavior, a run with:
- 100 sales orders took ~13 minutes.
- 1000 sales orders took ~66 minutes
⇒ ~900 invoices/hour, far below what we had before.
- We see a recurring pattern: after processing around 1000 invoices, performance drops further and some batch tasks go into Error.
- It looks like BATCHAOS2 stops processing work or becomes ineffective at that point.
- While the batch is running:
- CPU on the AOS/BATCHAOS servers is relatively low.
- But the “Memory Available (MBytes)” metric for BATCHAOS in LCS drops close to 0 MB at certain times → strong memory pressure.
- To keep invoicing moving at all, we had to split the workload into multiple smaller batch jobs (max. 1000 orders each) and run them sequentially, but throughput is still very poor.
Around the time we noticed the problem, there was also a CREATE INDEX on a custom Table running in the database (auto-tuning / DTA style _dta_index_…).
This index creation has since completed and is no longer running or blocking, but the slow invoice posting remains, so we don’t see the index as the current root cause anymore.
Question to the community
Has anyone seen similar behavior in PROD, where:
- Sales invoice posting becomes extremely slow,
- BATCHAOS memory drops close to zero while CPU stays low,
- and after ~1000 invoices tasks start failing / one BAOS instance seems to “stop working”?
- Known issues / hotfixes related to Sales invoice posting, batch framework or memory leaks,
- typical root causes in such scenarios (AOS topology, batch configuration, database/index problems),
- or concrete checks we should run (specific traces, performance counters, SQL traces, etc.)
Thanks in advance for any insights!
Best regards,
Stefan


Report
All responses (
Answers (