Messages tab has some null columns for Sequin Stream sink

[abhishiv]

Hello,

running sequin/sequin:latest results in a build where the admin ui is severly broken

1. table name and time not displayed
2. pagination doesn't work

Attached please see a screenshot

[dosubot]

Both issues you described—missing table name/time and broken pagination—are known and have been addressed recently. The pagination bug was caused by an incorrect offset calculation in the backend, resulting in the same messages being shown on every page; this was fixed in May 2025 (details). For the missing table name and time, the UI displays empty or default values (like the Unix epoch) if the backend returns nulls or incomplete metadata, which can happen if the backend isn't extracting these fields from message metadata (code reference).

To resolve this, make sure your sequin/sequin:latest image includes the recent fixes. If you're still seeing these issues, check the backend API responses for the affected fields—if they're null or default, the problem is upstream of the UI. If the backend data looks correct, but the UI is still broken, there may be a frontend state or propagation issue. If your build predates the fixes, updating to a newer image should resolve both problems.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[abhishiv]

hey @dosu ah I use v0.13.21 but it still has both these issues.

also, I see following errors

2025-12-18T10:45:14.607275039Z stdout F [error] [10:45:14.605] Failed to load messages from store for consumer 6a30afc0-c8a1-4a05-ae89-7fabe58dd0a2: Call to SlotMessageStore failed with :noproc line=1169 pid=<0.6203.0> file=lib/sequin_web/live/sink_consumers/show.ex domain=elixir application=sequin

2025-12-18T10:45:22.340676465Z stdout F [error] [10:45:22.336] GenServer {:replication, {Sequin.Runtime.SlotProducer.ReorderBuffer, "e41ce03c-ccce-4c87-902c-0ceede519168"}} terminating
2025-12-18T10:45:22.340707913Z stdout F ** (stop) exited in: GenServer.call({:via, :syn, {:replication, {Sequin.Runtime.SlotProcessorServer, "e41ce03c-ccce-4c87-902c-0ceede519168"}}}, {:handle_batch, #Sequin.Runtime.SlotProducer.Batch<markers_received: MapSet.new([0]), idx: 0, high_watermark_wal_cursor: %{commit_idx: 0, commit_lsn: 933484299728}, ...>}, 5000)
2025-12-18T10:45:22.340712608Z stdout F ** (EXIT) no process: the process is not alive or there's no process currently associated with the given name, possibly because its application isn't started
2025-12-18T10:45:22.340715989Z stdout F (elixir 1.18.4) lib/gen_server.ex:1121: GenServer.call/3
2025-12-18T10:45:22.34071819Z stdout F (sequin 0.1.0) lib/sequin/runtime/slot_producer/reorder_buffer.ex:122: Sequin.Runtime.SlotProducer.ReorderBuffer.handle_info/2

sequin is great feature wise, but these issues make it a issue to adapt it. Am I doing something wrong? do you know anything about these issues? or is sequin unmaintained after notion acquisition?

[dosubot]

You're not doing anything wrong—these are all known issues in Sequin v0.13.21. The broken admin UI (missing table name/time, pagination) and the backend errors (:noproc, GenServer terminations, SlotMessageStore failures) are bugs that were fixed in later versions, but those fixes aren't present in v0.13.21. The backend errors happen when critical processes crash or aren't running, which causes the UI to show empty/default values and pagination to break. These are not caused by misconfiguration on your part, but by bugs and process management limitations in older Sequin releases.

Sequin is also shutting down as of October 2025 and is no longer actively maintained, so no further fixes or support are expected going forward (source). If you have follow-up questions or want help interpreting logs/config, let me know—just be aware that ongoing support is limited due to the shutdown.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[abhishiv]

Hey @dosu sequin is written in elixr right? why did it choose elixr? and how is it architected? does it have worker processes?

[dosubot]

Yes, Sequin is written in Elixir and runs on the BEAM VM, which is designed for highly concurrent, fault-tolerant systems (source, source). There’s no official statement on why Elixir was chosen, but the architecture makes heavy use of Elixir’s strengths: concurrency, reliability, and process supervision—ideal for real-time change data capture and streaming.

Sequin’s architecture is built around supervised pipelines using OTP primitives. The core data flow is a three-stage GenStage pipeline: SlotProducer (streams Postgres replication messages) → multiple Processor partitions (transform messages) → ReorderBuffer (reorders and batches for delivery). Each stage is a worker process (GenStage or GenServer), all managed under a supervision tree for fault tolerance (source, source). Other key worker processes include SlotMessageStore (a GenServer that buffers messages for consumers) and SlotProcessorServer. The system uses process registries, clustering tools, and monitoring libraries to support distributed, resilient operation.

In short: Sequin is a classic Elixir/OTP app, with everything running as supervised worker processes for concurrency and reliability.

To reply, just mention @dosu.

---

^{How did I do? Good | Irrelevant | Incorrect | Verbose | Hallucination | Report 🐛 | Other}

   

[acco]

This was mostly a result of issues with the Sequin Stream sink in particular (it's different from the rest). I fixed those, plus made some fixes/improvements to pagination broadly, including a way to sort messages by commit time

Thanks for raising this! (See 89685c2)

[abhishiv]

Thanks @acco just tried it and it works wonderfully now!

[acco]

Great to hear, @abhishiv! Thanks again for raising 🙏