You picked the topic. Then it filled up.

Understand it one step at a time

The short runs these in order in about 32 seconds. Here they are written out — pick any step to jump the short straight to it.

1. 1

   You picked a topic. Good.

   Four teams needed the same event, so you created a topic and four subscriptions. Correct call.

   orders · topic · 4 subscriptions · healthy

2. 2

   Six months later

   The publisher starts throwing QuotaExceededException. Nothing about the publisher changed.

   send failed · topic is full

3. 3

   The $Default rule

   You added a filter to billing. You did not remove the rule Azure created for you, so it still matches everything.

   trap 1 · $Default TrueFilter still present

4. 4

   One idle subscription fills it

   A topic keeps a message until every subscription has completed it. legacy-export stopped reading in March.

   trap 2 · legacy-export · 4.1M messages held

5. 5

   N subs, N dead-letter queues

   Every subscription has its own DLQ. You set up one alert, years ago, on the queue this replaced.

   trap 3 · 4 DLQs · 1 alert

6. 6

   Ordering caps your fan-out

   Need FIFO? Enable sessions — per subscription. Now it is one consumer per session, and throughput drops.

   trap 4 · sessions · one consumer each

7. 7

   You can't convert a queue

   There is no toggle. A queue becomes a topic by creating a new entity and repointing every producer.

   trap 5 · no in-place migration

8. 8

   Five settings prevent all of it

   None of these are code. Remove the default rule, set TTLs, auto-delete idle subs, alert on every DLQ.

   five settings · five minutes

9. 9

   Queue or topic, honestly

   A queue is less to run. A topic is still the right default — as long as somebody owns the subscriptions.

   one destination · or four to maintain

10. 10

    A topic is a queue with homework

    The fan-out is free. The subscriptions are not — every one of them is a thing somebody has to own.

    Azure Service Bus · topics