langchain-ai/langsmith-trace-scripts
1
0
Fork 0
mirror of https://github.com/langchain-ai/langsmith-trace-scripts.git synced 2026-07-01 20:04:03 -04:00
Code Issues Packages Projects Releases Wiki Activity

Fix upload: Regenerate run IDs to avoid conflicts

Browse Source 
**Problem:** When uploading traces with run IDs that already exist in
LangSmith (e.g., from a previous upload), the API returns 409 conflict
and the runs don't get added to the new project. The script incorrectly
treated this as success.

**Solution:**
- Regenerate all run IDs before upload to ensure uniqueness
- Build ID mapping to preserve parent-child relationships
- Remove 409 handling since conflicts should no longer occur
- Add note about workspace determination via API key

This allows traces to be uploaded multiple times to different projects
without conflicts.

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude Sonnet 4.5 <noreply@anthropic.com>
This commit is contained in:
Stephen Chu
2025-12-23 21:37:23 -05:00
parent c7af2e379d
commit 3eb2b29924
2 changed files with 36 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
README.md
+3 -1
View File
@@ -32,6 +32,8 @@ export LANGSMITH_API_KEY='your-api-key'
The script will create a new tracing project with the specified name. You can delete the project later if needed.
**Note:** The trace will be uploaded to the workspace associated with your API key. Make sure you're using the correct API key for the intended workspace.
## Requirements
- `bash`
@@ -80,7 +82,7 @@ Upload scrubbed trace to a LangSmith tracing project.
./upload_trace.sh <trace_file> <project_name>
```
Creates a new tracing project with the specified name and uploads all runs to it.
Creates a new tracing project with the specified name and uploads all runs to it. Run IDs are automatically regenerated to avoid conflicts while preserving parent-child relationships.
## Complete Example
upload_trace.sh
+33 -6
View File
@@ -88,13 +88,43 @@ echo ""
UPLOADED=0
FAILED=0
TEMP_RUNS=$(mktemp)
trap "rm -f $TEMP_RUNS" EXIT
TEMP_MAPPING=$(mktemp)
trap "rm -f $TEMP_RUNS $TEMP_MAPPING" EXIT
# Build ID mapping (old_id -> new_id) to preserve parent-child relationships
jq -c 'sort_by(.dotted_order)[]' "$TRACE_FILE" > "$TEMP_RUNS"
echo "{}" > "$TEMP_MAPPING"
while read -r run; do
OLD_ID=$(echo "$run" | jq -r '.id')
NEW_ID=$(uuidgen | tr '[:upper:]' '[:lower:]')
TEMP_MAP=$(jq --arg old "$OLD_ID" --arg new "$NEW_ID" '.[$old] = $new' "$TEMP_MAPPING")
echo "$TEMP_MAP" > "$TEMP_MAPPING"
done < "$TEMP_RUNS"
# Apply mapping and upload
jq -c 'sort_by(.dotted_order)[]' "$TRACE_FILE" > "$TEMP_RUNS"
while read -r run; do
# Add session_id
run=$(echo "$run" | jq --arg sid "$SESSION_ID" '.session_id = $sid')
OLD_ID=$(echo "$run" | jq -r '.id')
OLD_PARENT=$(echo "$run" | jq -r '.parent_run_id // ""')
# Get new IDs from mapping
NEW_RUN_ID=$(jq -r --arg id "$OLD_ID" '.[$id]' "$TEMP_MAPPING")
NEW_PARENT_ID=$(if [ -n "$OLD_PARENT" ] && [ "$OLD_PARENT" != "null" ]; then
jq -r --arg id "$OLD_PARENT" '.[$id] // ""' "$TEMP_MAPPING"
else
echo ""
fi)
# Update run with new IDs and session
if [ -n "$NEW_PARENT_ID" ]; then
run=$(echo "$run" | jq --arg sid "$SESSION_ID" --arg rid "$NEW_RUN_ID" --arg pid "$NEW_PARENT_ID" \
'.session_id = $sid | .id = $rid | .parent_run_id = $pid')
else
run=$(echo "$run" | jq --arg sid "$SESSION_ID" --arg rid "$NEW_RUN_ID" \
'.session_id = $sid | .id = $rid | del(.parent_run_id)')
fi
RUN_ID=$(echo "$run" | jq -r '.id')
RUN_NAME=$(echo "$run" | jq -r '.name')
@@ -109,9 +139,6 @@ while read -r run; do
if [ "$HTTP_CODE" = "200" ] || [ "$HTTP_CODE" = "201" ] || [ "$HTTP_CODE" = "202" ]; then
UPLOADED=$((UPLOADED + 1))
echo " ✓ [$UPLOADED/$TOTAL] $RUN_NAME"
elif [ "$HTTP_CODE" = "409" ]; then
UPLOADED=$((UPLOADED + 1))
echo " ⊘ [$UPLOADED/$TOTAL] $RUN_NAME (already exists)"
else
FAILED=$((FAILED + 1))
echo " ✗ Failed: $RUN_NAME (HTTP $HTTP_CODE)"