# Handling Network Outages for Kafka

**URL:** <https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079>\
**Category:** Ignition 8.3 Early Access\
**Created:** [August 21, 2025, 11:11pm UTC](https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079 "2025-08-21T23:11:36Z")\
**Posts on this page:** 6\
**Page:** 1

<div class="post-metadata">

**Author:** ![achou](https://avatars.discourse-cdn.com/v4/letter/a/a6a055/32.png) [@achou](https://forum.inductiveautomation.com/u/achou)\
**Post date:** [August 21, 2025, 11:11pm UTC](https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079/1 "2025-08-21T23:11:36Z")

</div>

Very excited for the addition of Kafka to Ignition, wanted to ask if anyone’s found anything regarding how Ignition’s implementation of Kafka will support network outages. Specifically, Kafka producers have in memory buffers for storing messages until they’re successfully emitted, but haven’t seen any docs around whether this is the case in Ignition. Also unsure how or if Store & Forward plays into as well or is duplicative in function.

---

<div class="post-metadata">

**Author:** ![mrojas](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/mrojas/32/6394_2.png) [@mrojas](https://forum.inductiveautomation.com/u/mrojas)\
**Post date:** [August 22, 2025, 3:48pm UTC](https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079/2 "2025-08-22T15:48:15Z")

</div>

The Kafka producers in Ignition will definitely batch records using it’s internal memory buffer as you mentioned. We don’t use Store and Forward in Event Streams yet, but we do plan on incorporating it in the future.

---

<div class="post-metadata">

**Author:** ![Jonathan\_Meredith](https://avatars.discourse-cdn.com/v4/letter/j/a183cd/32.png) [@Jonathan\_Meredith](https://forum.inductiveautomation.com/u/Jonathan_Meredith)\
**Post date:** [August 22, 2025, 4:47pm UTC](https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079/3 "2025-08-22T16:47:12Z")

</div>

On gateway restart, are the unsent events stored somewhere and then restored?

---

<div class="post-metadata">

**Author:** ![achou](https://avatars.discourse-cdn.com/v4/letter/a/a6a055/32.png) [@achou](https://forum.inductiveautomation.com/u/achou)\
**Post date:** [August 22, 2025, 5:42pm UTC](https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079/4 "2025-08-22T17:42:13Z")

</div>

Appreciate the info! Just curious but is there any information about how large the buffer will be/is that something that can be configured? Just want to understand limits to estimate how long of an outage our system would handle

---

<div class="post-metadata">

**Author:** ![mrojas](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/mrojas/32/6394_2.png) [@mrojas](https://forum.inductiveautomation.com/u/mrojas)\
**Post date:** [August 22, 2025, 7:57pm UTC](https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079/5 "2025-08-22T19:57:46Z")

</div>

No, unfortunately they’re gone. We do plan on adding some sort of persistence eventually, probably through S&F.

---

<div class="post-metadata">

**Author:** ![mrojas](https://sea1.discourse-cdn.com/inductiveautomation/user_avatar/forum.inductiveautomation.com/mrojas/32/6394_2.png) [@mrojas](https://forum.inductiveautomation.com/u/mrojas)\
**Post date:** [August 22, 2025, 8:00pm UTC](https://forum.inductiveautomation.com/t/handling-network-outages-for-kafka/110079/6 "2025-08-22T20:00:10Z")

</div>

The default memory buffer for Kafka producers is 32MB, but this can be adjusted through the Kafka Connector settings producer props. `buffer.memory` is the property you’d have to adjust.