Upgrade from 1.10.0 to 1.11.0 consistently fail with h2 protocol error

TL;DR

Fix: Maintainer timvisee's guidance: with no incoming writes, upgrade each node in isolation (cluster comms not needed), then use the --reinit flag (added in 1.12.2 via PR #(5265) on the final upgrade; always snapshot/back up the cluster first. A community member posted a verified step-by-step upgrade path in the thread.

The error

Upgrading clusters 1.10.0 -> 1.11.x consistently breaks cluster communication with 'h2 protocol error: http2 error: stream error received: frame with invalid size', rendering many operations non-functional.

Fix

1. Maintainer timvisee's guidance: with no incoming writes, upgrade each node in isolation (cluster comms not needed), then use the --reinit flag (added in 1.12.2 via PR #(5265) on the final upgrade; always snapshot/back up the cluster first.
2. A community member posted a verified step-by-step upgrade path in the thread.

Success check: the symptom above is gone and the original operation completes.

When to use

• You hit this exact symptom in qdrant.
• You want the verified fix without digging through the thread.

When not to use

• A different error in qdrant, even a similar-looking one.
• General qdrant setup or credential questions.

Compatibility

• Qdrant (qdrant/qdrant)

Variant phrasings

• How to fix "upgrade from 1.10.0 to 1.11.0 consistently fail with h2 protocol error" in qdrant
• qdrant: upgrade from 1.10.0 to 1.11.0 consistently fail with h2 protocol error
• upgrade from 1.10.0 to 1.11.0 consistently fail with h2 protocol problem

Source

• https://github.com/qdrant/qdrant/issues/6541

In gh:qdrant/qdrant#6541 (https://github.com/qdrant/qdrant/issues/6541), the question was: Upgrading clusters 1.10.0 -> 1.11.x consistently breaks cluster communication with 'h2 protocol error: http2 error: stream error received: frame with invalid size', rendering many operations non-functional. The verified resolution: Maintainer timvisee's guidance: with no incoming writes, upgrade each node in isolation (cluster comms not needed), then use the --reinit flag (added in 1.12.2 via PR #5265) on the final upgrade; always snapshot/back up the cluster first. A community member posted a verified step-by-step upgrade path in the thread.