Conversation
columnFloat: setprecision(9) for IEEE 754 single (was default 6) columnDouble: setprecision(17) for IEEE 754 double (was default 6) Fixes BINARY_DOUBLE truncation, e.g. 65536.125 was output as 65536.1.
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (5)
💤 Files with no reviewable changes (4)
📝 WalkthroughWalkthroughAdds Debezium twin-test infrastructure (Oracle, OLR, Debezium LogMiner, receiver, orchestration and comparison scripts) and tweaks JSON numeric formatting precision in Changes
Sequence Diagram(s)mermaid Orchestrator->>Oracle: provision (01-setup.sh, 02-debezium-setup.sh) Estimated code review effort🎯 4 (Complex) | ⏱️ ~45 minutes Poem
🚥 Pre-merge checks | ✅ 2 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (2 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 5
🧹 Nitpick comments (2)
tests/scripts/compare-debezium.py (1)
67-84: Edge case in relative tolerance: asymmetric comparison may miss mismatches.The relative tolerance check at line 80 only considers
abs(fa)as the denominator. Iffais very small (e.g.,0.000001) butfbis much larger (e.g.,1.0), the conditionabs(fa - fb) / abs(fa)would be very large and fail as expected. However, if the values are swapped in the comparison (due to LogMiner vs OLR ordering), the behavior differs.More robust approach: use the larger absolute value as the reference:
♻️ Suggested fix for symmetric relative tolerance
# Numeric tolerance try: fa, fb = float(a), float(b) if fa == fb: return True - if fa != 0 and abs(fa - fb) / abs(fa) < 1e-6: + max_abs = max(abs(fa), abs(fb)) + if max_abs != 0 and abs(fa - fb) / max_abs < 1e-6: return True except (ValueError, TypeError): pass🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@tests/scripts/compare-debezium.py` around lines 67 - 84, The relative-tolerance check in values_match currently divides by abs(fa) which is asymmetric; update values_match to compute fa, fb as floats then use a symmetric denominator like max(abs(fa), abs(fb)) (and treat denominator == 0 as equal) when evaluating abs(fa - fb) / denominator < 1e-6 so that comparisons are order-independent; keep the existing exact-equality and None checks and still catch ValueError/TypeError around the float conversion.tests/scripts/run-debezium.sh (1)
120-145: Wait loops silently succeed on timeout, potentially masking initialization failures.The loops at lines 122-127, 132-137, and 139-144 will silently continue if the expected log patterns never appear. This could lead to flaky or misleading test results when OLR or connectors fail to initialize properly.
Consider adding timeout detection similar to Stage 5's approach:
♻️ Suggested fix for OLR wait loop (apply similar pattern to connector waits)
# Wait for OLR to start processing redo logs echo " Waiting for OLR to initialize..." +OLR_READY=false for i in $(seq 1 60); do if docker logs dbz-olr 2>&1 | tail -5 | grep -q "processing redo log"; then + OLR_READY=true break fi sleep 2 done +if [[ "$OLR_READY" != "true" ]]; then + echo "WARNING: OLR may not have initialized properly (timeout after 120s)" >&2 +fi echo " OLR: ready"🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@tests/scripts/run-debezium.sh` around lines 120 - 145, The three wait-for loops that poll docker logs for containers dbz-olr, dbz-olr-adapter, and dbz-logminer should detect timeouts and fail loudly instead of silently continuing; after each loop (the one checking for "processing redo log" on dbz-olr, the one checking for "streaming client started successfully|Starting streaming" on dbz-olr-adapter, and the one checking for "Starting streaming" on dbz-logminer) add an if that re-checks the grep condition and, if not matched, echo an error with context, dump the recent docker logs for that container, and exit 1 so the script fails fast on initialization errors. Ensure you reference the exact container names (dbz-olr, dbz-olr-adapter, dbz-logminer) and log patterns when implementing the timeout checks.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@tests/debezium/docker-compose.yaml`:
- Around line 32-47: Add a healthcheck to the olr service that verifies
OpenLogReplicator is listening on port 5000 (e.g., a TCP or HTTP check against
port 5000) and then gate the Debezium OLR adapter container (the service using
container_name dbz-olr / service name that runs the adapter) on olr’s readiness
by adding a depends_on entry pointing to olr with condition: service_healthy;
update the olr service’s healthcheck parameters (test, interval, timeout,
retries) so the compose wait is robust and Debezium won’t attempt to connect
before port 5000 is ready.
- Around line 20-31: The Debezium services currently only wait for the receiver
container to start, not to be ready, causing race conditions; add a Docker
healthcheck to the receiver service (the service named "receiver" running
command ["python3", "/app/debezium-receiver.py"] and exposing port 8080) that
probes the HTTP endpoint (e.g., curl http://localhost:8080/ or /health with a
short timeout and retries) and then update the Debezium services' depends_on
entries to require service_healthy for "receiver" so compose --wait will wait
for the receiver's healthcheck to pass before starting the Debezium services.
In `@tests/debezium/Makefile`:
- Around line 10-14: The download-ojdbc target currently treats any existing
lib/ojdbc8.jar as valid and can leave a zero-byte/truncated file; change the
recipe to download to a temporary file (e.g., lib/ojdbc8.jar.tmp or using
mktemp), verify wget succeeded, then atomically mv the temp file to
lib/ojdbc8.jar; ensure the lib directory is created first, remove the temp file
on failure, and still skip download when a valid lib/ojdbc8.jar exists;
reference the download-ojdbc target and the OJDBC_VERSION/OJDBC_URL variables
and the lib/ojdbc8.jar destination when making these changes.
In `@tests/debezium/oracle-init/01-setup.sh`:
- Around line 5-15: The SQL*Plus block lacks error handling and sets ALTER
SYSTEM SET log_archive_dest_1 before the target directory exists; add WHENEVER
SQLERROR EXIT SQL.SQLCODE and WHENEVER OSERROR EXIT FAILURE at the top of the
sqlplus session to abort on errors, move the HOST mkdir -p
/opt/oracle/oradata/XE/archive so it runs before ALTER SYSTEM SET
log_archive_dest_1='LOCATION=...' (so the path exists when Oracle validates it),
and ensure the sqlplus invocation checks the exit code so the init script fails
fast if either the mkdir or ALTER SYSTEM call fails.
In `@tests/debezium/oracle-init/02-debezium-setup.sh`:
- Line 5: Add SQL*Plus error handling at the start of the heredoc launched by
"sqlplus -S / as sysdba <<'SQL'" so the script stops on the first SQL or OS
error; insert the two directives ("WHENEVER SQLERROR ..." and "WHENEVER OSERROR
...") immediately after the heredoc begins to force an immediate exit on
failures (so bad GRANTs for c##dbzuser halt provisioning).
---
Nitpick comments:
In `@tests/scripts/compare-debezium.py`:
- Around line 67-84: The relative-tolerance check in values_match currently
divides by abs(fa) which is asymmetric; update values_match to compute fa, fb as
floats then use a symmetric denominator like max(abs(fa), abs(fb)) (and treat
denominator == 0 as equal) when evaluating abs(fa - fb) / denominator < 1e-6 so
that comparisons are order-independent; keep the existing exact-equality and
None checks and still catch ValueError/TypeError around the float conversion.
In `@tests/scripts/run-debezium.sh`:
- Around line 120-145: The three wait-for loops that poll docker logs for
containers dbz-olr, dbz-olr-adapter, and dbz-logminer should detect timeouts and
fail loudly instead of silently continuing; after each loop (the one checking
for "processing redo log" on dbz-olr, the one checking for "streaming client
started successfully|Starting streaming" on dbz-olr-adapter, and the one
checking for "Starting streaming" on dbz-logminer) add an if that re-checks the
grep condition and, if not matched, echo an error with context, dump the recent
docker logs for that container, and exit 1 so the script fails fast on
initialization errors. Ensure you reference the exact container names (dbz-olr,
dbz-olr-adapter, dbz-logminer) and log patterns when implementing the timeout
checks.
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 4a53acd2-80ca-4698-8383-da35271a4398
📒 Files selected for processing (12)
src/builder/BuilderJson.cpptests/debezium/.gitignoretests/debezium/Makefiletests/debezium/config/application-logminer.propertiestests/debezium/config/application-olr.propertiestests/debezium/config/olr-config.jsontests/debezium/docker-compose.yamltests/debezium/oracle-init/01-setup.shtests/debezium/oracle-init/02-debezium-setup.shtests/scripts/compare-debezium.pytests/scripts/debezium-receiver.pytests/scripts/run-debezium.sh
| receiver: | ||
| image: python:3.12-slim | ||
| container_name: dbz-receiver | ||
| command: ["python3", "/app/debezium-receiver.py"] | ||
| ports: | ||
| - "8080:8080" | ||
| environment: | ||
| OUTPUT_DIR: /app/output | ||
| volumes: | ||
| - ../scripts/debezium-receiver.py:/app/debezium-receiver.py:ro | ||
| - ./output:/app/output | ||
|
|
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# First, let's get the full docker-compose.yaml file
echo "=== Full docker-compose.yaml file ==="
cat -n tests/debezium/docker-compose.yamlRepository: rophy/OpenLogReplicator
Length of output: 2896
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "=== Checking Makefile for docker compose commands ==="
cat -n tests/debezium/Makefile | grep -A 2 -B 2 "docker compose"
echo
echo "=== Checking debezium-receiver.py for port 8080 setup ==="
if [ -f scripts/debezium-receiver.py ]; then
head -100 scripts/debezium-receiver.py | grep -A 5 -B 5 "8080\|listen\|port" || echo "Pattern not found in first 100 lines"
fiRepository: rophy/OpenLogReplicator
Length of output: 331
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "=== Full debezium-receiver.py file ==="
cat -n scripts/debezium-receiver.pyRepository: rophy/OpenLogReplicator
Length of output: 183
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "=== Searching for debezium-receiver.py ==="
fd -t f "debezium-receiver.py"
echo
echo "=== Searching for receiver scripts ==="
fd -t f -i "receiver" | head -20Repository: rophy/OpenLogReplicator
Length of output: 219
🏁 Script executed:
#!/bin/bash
set -euo pipefail
echo "=== Full debezium-receiver.py file ==="
cat -n tests/scripts/debezium-receiver.pyRepository: rophy/OpenLogReplicator
Length of output: 6380
Gate the Debezium services on receiver readiness, not just startup.
make up uses docker compose up --wait, but receiver has no healthcheck and both Debezium services only depend on it being "started". If the Python HTTP server has not bound :8080 yet, the first posts can race and make the twin-test flaky.
🛠️ Suggested fix
receiver:
image: python:3.12-slim
container_name: dbz-receiver
command: ["python3", "/app/debezium-receiver.py"]
ports:
- "8080:8080"
environment:
OUTPUT_DIR: /app/output
volumes:
- ../scripts/debezium-receiver.py:/app/debezium-receiver.py:ro
- ./output:/app/output
+ healthcheck:
+ test: ["CMD", "python3", "-c", "import socket; socket.create_connection(('127.0.0.1', 8080), 2).close()"]
+ interval: 5s
+ timeout: 3s
+ retries: 20
dbz-logminer:
image: quay.io/debezium/server:3.0
container_name: dbz-logminer
depends_on:
oracle:
condition: service_healthy
receiver:
- condition: service_started
+ condition: service_healthy
@@
dbz-olr:
image: quay.io/debezium/server:3.0
container_name: dbz-olr-adapter
depends_on:
olr:
condition: service_started
receiver:
- condition: service_started
+ condition: service_healthy📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| receiver: | |
| image: python:3.12-slim | |
| container_name: dbz-receiver | |
| command: ["python3", "/app/debezium-receiver.py"] | |
| ports: | |
| - "8080:8080" | |
| environment: | |
| OUTPUT_DIR: /app/output | |
| volumes: | |
| - ../scripts/debezium-receiver.py:/app/debezium-receiver.py:ro | |
| - ./output:/app/output | |
| receiver: | |
| image: python:3.12-slim | |
| container_name: dbz-receiver | |
| command: ["python3", "/app/debezium-receiver.py"] | |
| ports: | |
| - "8080:8080" | |
| environment: | |
| OUTPUT_DIR: /app/output | |
| volumes: | |
| - ../scripts/debezium-receiver.py:/app/debezium-receiver.py:ro | |
| - ./output:/app/output | |
| healthcheck: | |
| test: ["CMD", "python3", "-c", "import socket; socket.create_connection(('127.0.0.1', 8080), 2).close()"] | |
| interval: 5s | |
| timeout: 3s | |
| retries: 20 |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@tests/debezium/docker-compose.yaml` around lines 20 - 31, The Debezium
services currently only wait for the receiver container to start, not to be
ready, causing race conditions; add a Docker healthcheck to the receiver service
(the service named "receiver" running command ["python3",
"/app/debezium-receiver.py"] and exposing port 8080) that probes the HTTP
endpoint (e.g., curl http://localhost:8080/ or /health with a short timeout and
retries) and then update the Debezium services' depends_on entries to require
service_healthy for "receiver" so compose --wait will wait for the receiver's
healthcheck to pass before starting the Debezium services.
| olr: | ||
| image: olr-dev:${OLR_IMAGE_TAG:-latest} | ||
| container_name: dbz-olr | ||
| entrypoint: ["/bin/bash", "-c", "mkdir -p /olr-data/checkpoint && exec /opt/OpenLogReplicator/OpenLogReplicator \"$@\"", "--"] | ||
| command: ["-r", "-f", "/config/olr-config.json"] | ||
| working_dir: /olr-data | ||
| group_add: | ||
| - "54321" # oinstall group — read access to Oracle redo logs | ||
| depends_on: | ||
| oracle: | ||
| condition: service_healthy | ||
| tmpfs: | ||
| - /olr-data:uid=1000,gid=54322 | ||
| volumes: | ||
| - ./config/olr-config.json:/config/olr-config.json:ro | ||
| - oracle-data:/opt/oracle/oradata:ro |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Check if file exists and show relevant lines
file="tests/debezium/docker-compose.yaml"
if [ -f "$file" ]; then
echo "=== File exists ==="
echo
echo "=== Lines 32-47 (olr service) ==="
head -47 "$file" | tail -16 | cat -n
echo
echo "=== Lines 65-70 (dbz-olr and its depends_on) ==="
wc_result=$(wc -l < "$file")
echo "File has $wc_result total lines"
if [ "$wc_result" -ge 70 ]; then
head -70 "$file" | tail -10 | cat -n
else
echo "File is shorter than expected"
tail -20 "$file" | cat -n
fi
else
echo "File not found: $file"
exit 1
fiRepository: rophy/OpenLogReplicator
Length of output: 1192
🏁 Script executed:
#!/bin/bash
# List all files in the tests directory to understand structure
if [ -d "tests" ]; then
echo "=== Contents of tests directory ==="
ls -la tests/ | head -20
echo
if [ -d "tests/debezium" ]; then
echo "=== Contents of tests/debezium directory ==="
ls -la tests/debezium/ | head -20
fi
else
echo "tests directory not found"
fiRepository: rophy/OpenLogReplicator
Length of output: 1183
🏁 Script executed:
#!/bin/bash
# Search for port 5000 and OpenLogReplicator port references
echo "=== Searching for port 5000 references ==="
rg "5000" tests/debezium/
echo
echo "=== Checking OLR config file ==="
if [ -f "tests/debezium/config/olr-config.json" ]; then
cat tests/debezium/config/olr-config.json | head -50
fi
echo
echo "=== Searching for any healthcheck patterns in the repo ==="
rg "healthcheck" . --type yaml | head -10Repository: rophy/OpenLogReplicator
Length of output: 1628
🏁 Script executed:
#!/bin/bash
# Search for OpenLogReplicator listener or networking configuration
echo "=== Searching for listener config in codebase ==="
rg -i "listener|port.*5000|5000.*port" . --type json --type yaml | head -20Repository: rophy/OpenLogReplicator
Length of output: 50380
Add healthcheck to olr service and gate dbz-olr on OLR readiness.
The current setup waits only for the OLR container to start, not for OpenLogReplicator to bind port 5000. This creates a race condition where Debezium's OLR adapter may attempt to connect before the service is listening.
Port 5000 is confirmed as the OLR listener port in both the OLR config (uri: 0.0.0.0:5000) and Debezium properties (openlogreplicator.port=5000).
Suggested fix
olr:
image: olr-dev:${OLR_IMAGE_TAG:-latest}
container_name: dbz-olr
entrypoint: ["/bin/bash", "-c", "mkdir -p /olr-data/checkpoint && exec /opt/OpenLogReplicator/OpenLogReplicator \"$@\"", "--"]
command: ["-r", "-f", "/config/olr-config.json"]
working_dir: /olr-data
group_add:
- "54321" # oinstall group — read access to Oracle redo logs
depends_on:
oracle:
condition: service_healthy
tmpfs:
- /olr-data:uid=1000,gid=54322
volumes:
- ./config/olr-config.json:/config/olr-config.json:ro
- oracle-data:/opt/oracle/oradata:ro
+ healthcheck:
+ test: ["CMD", "bash", "-lc", "exec 3<>/dev/tcp/127.0.0.1/5000"]
+ interval: 5s
+ timeout: 3s
+ retries: 20
dbz-olr:
image: quay.io/debezium/server:3.0
container_name: dbz-olr-adapter
depends_on:
olr:
- condition: service_started
+ condition: service_healthy
receiver:
condition: service_started🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@tests/debezium/docker-compose.yaml` around lines 32 - 47, Add a healthcheck
to the olr service that verifies OpenLogReplicator is listening on port 5000
(e.g., a TCP or HTTP check against port 5000) and then gate the Debezium OLR
adapter container (the service using container_name dbz-olr / service name that
runs the adapter) on olr’s readiness by adding a depends_on entry pointing to
olr with condition: service_healthy; update the olr service’s healthcheck
parameters (test, interval, timeout, retries) so the compose wait is robust and
Debezium won’t attempt to connect before port 5000 is ready.
| download-ojdbc: | ||
| @mkdir -p lib | ||
| @test -f lib/ojdbc8.jar || \ | ||
| (echo "Downloading ojdbc8-$(OJDBC_VERSION).jar..." && \ | ||
| wget -q -O lib/ojdbc8.jar $(OJDBC_URL)) |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
file="tests/debezium/Makefile"
# First, check if the file exists
if [ ! -f "$file" ]; then
echo "File not found: $file"
exit 1
fi
echo "=== Full Makefile content ==="
cat -n "$file"Repository: rophy/OpenLogReplicator
Length of output: 1140
Download the JDBC jar atomically.
This recipe treats any existing lib/ojdbc8.jar as valid. A zero-byte or truncated file from an interrupted download will be cached and every later make up will silently reuse it.
🛠️ Suggested fix
download-ojdbc:
`@mkdir` -p lib
- `@test` -f lib/ojdbc8.jar || \
- (echo "Downloading ojdbc8-$(OJDBC_VERSION).jar..." && \
- wget -q -O lib/ojdbc8.jar $(OJDBC_URL))
+ `@test` -s lib/ojdbc8.jar || \
+ (tmp=$$(mktemp lib/ojdbc8.jar.XXXXXX) && \
+ trap 'rm -f "$$tmp"' EXIT && \
+ echo "Downloading ojdbc8-$(OJDBC_VERSION).jar..." && \
+ wget -q -O "$$tmp" $(OJDBC_URL) && \
+ mv "$$tmp" lib/ojdbc8.jar)📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| download-ojdbc: | |
| @mkdir -p lib | |
| @test -f lib/ojdbc8.jar || \ | |
| (echo "Downloading ojdbc8-$(OJDBC_VERSION).jar..." && \ | |
| wget -q -O lib/ojdbc8.jar $(OJDBC_URL)) | |
| download-ojdbc: | |
| `@mkdir` -p lib | |
| `@test` -s lib/ojdbc8.jar || \ | |
| (tmp=$$(mktemp lib/ojdbc8.jar.XXXXXX) && \ | |
| trap 'rm -f "$$tmp"' EXIT && \ | |
| echo "Downloading ojdbc8-$(OJDBC_VERSION).jar..." && \ | |
| wget -q -O "$$tmp" $(OJDBC_URL) && \ | |
| mv "$$tmp" lib/ojdbc8.jar) |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@tests/debezium/Makefile` around lines 10 - 14, The download-ojdbc target
currently treats any existing lib/ojdbc8.jar as valid and can leave a
zero-byte/truncated file; change the recipe to download to a temporary file
(e.g., lib/ojdbc8.jar.tmp or using mktemp), verify wget succeeded, then
atomically mv the temp file to lib/ojdbc8.jar; ensure the lib directory is
created first, remove the temp file on failure, and still skip download when a
valid lib/ojdbc8.jar exists; reference the download-ojdbc target and the
OJDBC_VERSION/OJDBC_URL variables and the lib/ojdbc8.jar destination when making
these changes.
| sqlplus -S / as sysdba <<'SQL' | ||
| SHUTDOWN IMMEDIATE; | ||
| STARTUP MOUNT; | ||
| ALTER DATABASE ARCHIVELOG; | ||
| ALTER DATABASE OPEN; | ||
| ALTER DATABASE ADD SUPPLEMENTAL LOG DATA; | ||
| ALTER SYSTEM SET db_recovery_file_dest_size=10G; | ||
|
|
||
| -- Archive to shared volume so OLR container can read them | ||
| ALTER SYSTEM SET log_archive_dest_1='LOCATION=/opt/oracle/oradata/XE/archive' SCOPE=BOTH; | ||
| HOST mkdir -p /opt/oracle/oradata/XE/archive |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Check if file exists and read lines 5-15
file="tests/debezium/oracle-init/01-setup.sh"
if [ ! -f "$file" ]; then
echo "File not found: $file"
exit 1
fi
echo "=== Full file content ==="
cat -n "$file"
echo
echo "=== Lines 5-15 specifically ==="
sed -n '5,15p' "$file" | cat -n
echo
echo "=== Checking for WHENEVER directives ==="
grep -n 'WHENEVER' "$file" || echo "No WHENEVER directives found"
echo
echo "=== Checking order of mkdir and ALTER SYSTEM SET ==="
grep -n 'mkdir\|ALTER SYSTEM SET log_archive_dest_1' "$file"Repository: rophy/OpenLogReplicator
Length of output: 1708
Add error handling and reorder the archive destination setup.
The script sets log_archive_dest_1 before the /opt/oracle/oradata/XE/archive directory exists, and lacks WHENEVER SQLERROR or WHENEVER OSERROR directives. If Oracle rejects the path, SQL*Plus continues silently and the init appears successful while archive logging is misconfigured.
🛠️ Suggested fix
-sqlplus -S / as sysdba <<'SQL'
+sqlplus -S / as sysdba <<'SQL'
+WHENEVER OSERROR EXIT FAILURE
+WHENEVER SQLERROR EXIT SQL.SQLCODE
SHUTDOWN IMMEDIATE;
STARTUP MOUNT;
ALTER DATABASE ARCHIVELOG;
ALTER DATABASE OPEN;
ALTER DATABASE ADD SUPPLEMENTAL LOG DATA;
ALTER SYSTEM SET db_recovery_file_dest_size=10G;
-- Archive to shared volume so OLR container can read them
-ALTER SYSTEM SET log_archive_dest_1='LOCATION=/opt/oracle/oradata/XE/archive' SCOPE=BOTH;
HOST mkdir -p /opt/oracle/oradata/XE/archive
+ALTER SYSTEM SET log_archive_dest_1='LOCATION=/opt/oracle/oradata/XE/archive' SCOPE=BOTH;📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| sqlplus -S / as sysdba <<'SQL' | |
| SHUTDOWN IMMEDIATE; | |
| STARTUP MOUNT; | |
| ALTER DATABASE ARCHIVELOG; | |
| ALTER DATABASE OPEN; | |
| ALTER DATABASE ADD SUPPLEMENTAL LOG DATA; | |
| ALTER SYSTEM SET db_recovery_file_dest_size=10G; | |
| -- Archive to shared volume so OLR container can read them | |
| ALTER SYSTEM SET log_archive_dest_1='LOCATION=/opt/oracle/oradata/XE/archive' SCOPE=BOTH; | |
| HOST mkdir -p /opt/oracle/oradata/XE/archive | |
| sqlplus -S / as sysdba <<'SQL' | |
| WHENEVER OSERROR EXIT FAILURE | |
| WHENEVER SQLERROR EXIT SQL.SQLCODE | |
| SHUTDOWN IMMEDIATE; | |
| STARTUP MOUNT; | |
| ALTER DATABASE ARCHIVELOG; | |
| ALTER DATABASE OPEN; | |
| ALTER DATABASE ADD SUPPLEMENTAL LOG DATA; | |
| ALTER SYSTEM SET db_recovery_file_dest_size=10G; | |
| -- Archive to shared volume so OLR container can read them | |
| HOST mkdir -p /opt/oracle/oradata/XE/archive | |
| ALTER SYSTEM SET log_archive_dest_1='LOCATION=/opt/oracle/oradata/XE/archive' SCOPE=BOTH; |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@tests/debezium/oracle-init/01-setup.sh` around lines 5 - 15, The SQL*Plus
block lacks error handling and sets ALTER SYSTEM SET log_archive_dest_1 before
the target directory exists; add WHENEVER SQLERROR EXIT SQL.SQLCODE and WHENEVER
OSERROR EXIT FAILURE at the top of the sqlplus session to abort on errors, move
the HOST mkdir -p /opt/oracle/oradata/XE/archive so it runs before ALTER SYSTEM
SET log_archive_dest_1='LOCATION=...' (so the path exists when Oracle validates
it), and ensure the sqlplus invocation checks the exit code so the init script
fails fast if either the mkdir or ALTER SYSTEM call fails.
| # Create Debezium user with LogMiner privileges + sentinel table. | ||
| # Runs after 01-setup.sh during gvenzl/oracle-xe container initialization. | ||
|
|
||
| sqlplus -S / as sysdba <<'SQL' |
There was a problem hiding this comment.
🧩 Analysis chain
🏁 Script executed:
#!/bin/bash
set -euo pipefail
# Check if the file exists
file="tests/debezium/oracle-init/02-debezium-setup.sh"
if [ ! -f "$file" ]; then
echo "File not found: $file"
exit 1
fi
echo "=== Checking for WHENEVER directives ==="
rg -n 'WHENEVER (SQLERROR|OSERROR)' "$file" || echo "No WHENEVER directives found"
echo
echo "=== SQL*Plus heredoc start (lines 1-20) ==="
head -20 "$file" | cat -n
echo
echo "=== Looking for sqlplus heredoc patterns ==="
rg -n "sqlplus" "$file"
echo
echo "=== File size ==="
wc -l "$file"Repository: rophy/OpenLogReplicator
Length of output: 1141
🏁 Script executed:
#!/bin/bash
# Check config files for c##dbzuser reference
echo "=== Checking config files for c##dbzuser ==="
find tests/debezium/config -name "application-*.properties" -type f | while read f; do
echo "File: $f"
rg 'c##dbzuser|dbzuser' "$f" || echo " No dbzuser reference found"
doneRepository: rophy/OpenLogReplicator
Length of output: 308
🏁 Script executed:
#!/bin/bash
# Get the complete SQL*Plus heredoc to understand the full scope
file="tests/debezium/oracle-init/02-debezium-setup.sh"
echo "=== Complete SQL*Plus heredoc ==="
sed -n '5,/^SQL$/p' "$file" | cat -n
echo
echo "=== Context: end of file (last 10 lines) ==="
tail -10 "$file" | cat -nRepository: rophy/OpenLogReplicator
Length of output: 4764
Stop this provisioning script on the first failed grant.
Both Debezium configs connect as c##dbzuser, so letting SQL*Plus continue past a bad grant here turns into much later connector failures. Add WHENEVER SQLERROR/OSERROR at the top of the heredoc.
🛠️ Suggested fix
-sqlplus -S / as sysdba <<'SQL'
+sqlplus -S / as sysdba <<'SQL'
+WHENEVER OSERROR EXIT FAILURE
+WHENEVER SQLERROR EXIT SQL.SQLCODE
-- Create common user for Debezium LogMiner access (CDB level)
CREATE USER c##dbzuser IDENTIFIED BY dbz📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
| sqlplus -S / as sysdba <<'SQL' | |
| sqlplus -S / as sysdba <<'SQL' | |
| WHENEVER OSERROR EXIT FAILURE | |
| WHENEVER SQLERROR EXIT SQL.SQLCODE | |
| -- Create common user for Debezium LogMiner access (CDB level) | |
| CREATE USER c##dbzuser IDENTIFIED BY dbz |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@tests/debezium/oracle-init/02-debezium-setup.sh` at line 5, Add SQL*Plus
error handling at the start of the heredoc launched by "sqlplus -S / as sysdba
<<'SQL'" so the script stops on the first SQL or OS error; insert the two
directives ("WHENEVER SQLERROR ..." and "WHENEVER OSERROR ...") immediately
after the heredoc begins to force an immediate exit on failures (so bad GRANTs
for c##dbzuser halt provisioning).
1 participant
Summary
Fixes issue #3 (Issue B): BINARY_DOUBLE and BINARY_FLOAT values truncated to 6 significant digits in JSON output.
columnFloat(): setstd::setprecision(9)(IEEE 754 single needs 9 digits for round-trip)columnDouble(): setstd::setprecision(17)(IEEE 754 double needs 17 digits for round-trip)Before:
65536.125→65536.1,3.14159265358979→3.14159After:
65536.125→65536.125,3.14159265358979→3.1415926535897931Test plan
COL_DOUBLEdiffsSummary by CodeRabbit
New Features
Improvements
Chores