Hi all,
I’ve run into a reproducible issue where module installation and module import (.rei3 files) fail on x86_64, while the exact same setup works fine on ARM64.
✅ Working environment
• Architecture: ARM64
• OS: Debian 13
• REI3 versions tested: 3.10.7 + 3.11.6
• Behavior:
• Installing apps from the repository works
• Importing .rei3 files works
• Local custom apps (tables, uploads) work
• No errors
❌ Failing environment
• Architecture: x86_64 (amd64)
• OS: Debian 13 (fresh install)
• REI3 versions tested: 3.10.7 + 3.11.6 (same result)
• DB: PostgreSQL 15.14 on a separate server (same DB backend used by ARM setup)
❗ Symptoms on x86_64
• Installing from the Repository hangs for a long time, then fails.
• Importing any .rei3 file also fails (same behavior).
• Random error messages during/after import:
could not finish module import, timeout: context deadline exceeded
failed to deallocate cached statement(s): conn closed
connection closed
(Errors vary per attempt.)
• Creating a custom app (schema + file upload) works perfectly on the same instance → DB and filesystem access are fine.
❗ Verified NOT to be the cause
• Not a DB issue (pgcrypto installed, permissions OK, same DB works for ARM)
• Not a TLS problem
• Not a repo problem (ARM instance installs modules at the same time)
• Not filesystem permissions (custom uploads work)
• Not a REI3 version issue (both 3.10 and 3.11 fail on x64)
Conclusion
Everything works on ARM, but module import (repo + file) consistently fails on x86_64, suggesting a platform-specific issue in the module import pipeline (ZIP extraction, JSON parsing, temp handling, timeouts, etc.).
Happy to provide logs, .rei3 files, or a test VM if needed.
Thanks!