Skip to content

Fix Addie performance metrics SQL and data storage #617

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
bokelley merged 1 commit into from
Jan 5, 2026
Merged

Fix Addie performance metrics SQL and data storage #617

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
2 changes: 2 additions & 0 deletions .changeset/eleven-rabbits-enter.md
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,2 @@
---
---
55 changes: 39 additions & 16 deletions server/src/addie/bolt-app.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -942,8 +942,11 @@ async function handleUserMessage({
}
}

// Build final response object if we used streaming
// Build final response object if we used streaming but didn't receive a 'done' event
// This shouldn't happen normally, but provides a fallback with logging
if (!response) {
logger.warn({ fullTextLength: fullText.length, toolsUsedCount: toolsUsed.length },
'Addie Bolt: Streaming completed without done event - using fallback response');
response = {
text: fullText,
tools_used: toolsUsed,
Expand All @@ -953,7 +956,8 @@ async function handleUserMessage({
duration_ms: 0,
sequence: i + 1,
})),
flagged: false,
flagged: true,
flag_reason: 'Streaming completed without done event',
};
}

Expand Down Expand Up @@ -2334,6 +2338,7 @@ async function handleReactionAdded({
'Addie Bolt: Received reaction on Addie message'
);

const startTime = Date.now();
const threadService = getThreadService();

// Build external ID to find the thread
Expand Down Expand Up @@ -2468,22 +2473,40 @@ async function handleReactionAdded({
logger.error({ error }, 'Addie Bolt: Failed to send reaction response');
}

// Log assistant response
await threadService.addMessage({
thread_id: thread.thread_id,
role: 'assistant',
content: response.text,
tools_used: response.tools_used,
tool_calls: response.tool_executions?.map(exec => ({
name: exec.tool_name,
input: exec.parameters,
result: exec.result,
})),
model: AddieModelConfig.chat,
});
// Log assistant response with performance data
try {
await threadService.addMessage({
thread_id: thread.thread_id,
role: 'assistant',
content: response.text,
tools_used: response.tools_used,
tool_calls: response.tool_executions?.map(exec => ({
name: exec.tool_name,
input: exec.parameters,
result: exec.result,
duration_ms: exec.duration_ms,
is_error: exec.is_error,
})),
model: AddieModelConfig.chat,
latency_ms: Date.now() - startTime,
tokens_input: response.usage?.input_tokens,
tokens_output: response.usage?.output_tokens,
timing: response.timing ? {
system_prompt_ms: response.timing.system_prompt_ms,
total_llm_ms: response.timing.total_llm_ms,
total_tool_ms: response.timing.total_tool_execution_ms,
iterations: response.timing.iterations,
} : undefined,
tokens_cache_creation: response.usage?.cache_creation_input_tokens,
tokens_cache_read: response.usage?.cache_read_input_tokens,
active_rule_ids: response.active_rule_ids,
});
} catch (error) {
logger.error({ error, threadId: thread.thread_id }, 'Addie Bolt: Failed to log reaction response');
}

logger.info(
{ threadId: thread.thread_id, reaction, isConfirmation: isConfirmationRequest },
{ threadId: thread.thread_id, reaction, isConfirmation: isConfirmationRequest, latencyMs: Date.now() - startTime },
'Addie Bolt: Processed reaction and responded'
);
}
58 changes: 29 additions & 29 deletions server/src/addie/thread-service.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -392,26 +392,26 @@ export class ThreadService {
input.thread_id,
input.role,
input.content,
input.content_sanitized || null,
input.tools_used || null,
input.content_sanitized ?? null,
input.tools_used ?? null,
input.tool_calls ? JSON.stringify(input.tool_calls) : null,
input.knowledge_ids || null,
input.model || null,
input.latency_ms || null,
input.tokens_input || null,
input.tokens_output || null,
input.flagged || false,
input.flag_reason || null,
input.knowledge_ids ?? null,
input.model ?? null,
input.latency_ms ?? null,
input.tokens_input ?? null,
input.tokens_output ?? null,
input.flagged ?? false,
input.flag_reason ?? null,
sequenceNumber,
input.timing?.system_prompt_ms || null,
input.timing?.total_llm_ms || null,
input.timing?.total_tool_ms || null,
input.timing?.iterations || null,
input.tokens_cache_creation || null,
input.tokens_cache_read || null,
input.active_rule_ids || null,
input.timing?.system_prompt_ms ?? null,
input.timing?.total_llm_ms ?? null,
input.timing?.total_tool_ms ?? null,
input.timing?.iterations ?? null,
input.tokens_cache_creation ?? null,
input.tokens_cache_read ?? null,
input.active_rule_ids ?? null,
input.router_decision ? JSON.stringify(input.router_decision) : null,
input.config_version_id || null,
input.config_version_id ?? null,
]
);

Expand Down Expand Up @@ -909,14 +909,14 @@ export class ThreadService {
`SELECT
COUNT(*) as total_messages,
COUNT(*) FILTER (WHERE role = 'assistant') as total_assistant_messages,
ROUND(AVG(latency_ms) FILTER (WHERE role = 'assistant')::numeric, 0) as avg_latency_ms,
ROUND(PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY latency_ms) FILTER (WHERE role = 'assistant')::numeric, 0) as p50_latency_ms,
ROUND(PERCENTILE_CONT(0.95) WITHIN GROUP (ORDER BY latency_ms) FILTER (WHERE role = 'assistant')::numeric, 0) as p95_latency_ms,
ROUND((AVG(latency_ms) FILTER (WHERE role = 'assistant'))::numeric, 0) as avg_latency_ms,
ROUND((PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY latency_ms) FILTER (WHERE role = 'assistant'))::numeric, 0) as p50_latency_ms,
ROUND((PERCENTILE_CONT(0.95) WITHIN GROUP (ORDER BY latency_ms) FILTER (WHERE role = 'assistant'))::numeric, 0) as p95_latency_ms,
MAX(latency_ms) FILTER (WHERE role = 'assistant') as max_latency_ms,
COALESCE(SUM(tokens_input), 0) as total_input_tokens,
COALESCE(SUM(tokens_output), 0) as total_output_tokens,
ROUND(AVG(tokens_input) FILTER (WHERE tokens_input IS NOT NULL)::numeric, 0) as avg_input_tokens,
ROUND(AVG(tokens_output) FILTER (WHERE tokens_output IS NOT NULL)::numeric, 0) as avg_output_tokens
ROUND((AVG(tokens_input) FILTER (WHERE tokens_input IS NOT NULL))::numeric, 0) as avg_input_tokens,
ROUND((AVG(tokens_output) FILTER (WHERE tokens_output IS NOT NULL))::numeric, 0) as avg_output_tokens
FROM addie_thread_messages
WHERE created_at > NOW() - make_interval(days => $1)`,
[days]
Expand Down Expand Up @@ -955,8 +955,8 @@ export class ThreadService {
`SELECT
COALESCE(model, 'unknown') as model,
COUNT(*) as count,
ROUND(AVG(latency_ms)::numeric, 0) as avg_latency_ms,
ROUND(PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY latency_ms)::numeric, 0) as p50_latency_ms,
ROUND((AVG(latency_ms))::numeric, 0) as avg_latency_ms,
ROUND((PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY latency_ms))::numeric, 0) as p50_latency_ms,
COALESCE(SUM(tokens_input), 0) as total_input_tokens,
COALESCE(SUM(tokens_output), 0) as total_output_tokens
FROM addie_thread_messages
Expand Down Expand Up @@ -987,9 +987,9 @@ export class ThreadService {
SELECT
tool->>'name' as tool_name,
COUNT(*) as call_count,
ROUND(AVG((tool->>'duration_ms')::numeric), 0) as avg_duration_ms,
ROUND(PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY (tool->>'duration_ms')::numeric), 0) as p50_duration_ms,
ROUND(PERCENTILE_CONT(0.95) WITHIN GROUP (ORDER BY (tool->>'duration_ms')::numeric), 0) as p95_duration_ms,
ROUND((AVG((tool->>'duration_ms')::numeric))::numeric, 0) as avg_duration_ms,
ROUND((PERCENTILE_CONT(0.5) WITHIN GROUP (ORDER BY (tool->>'duration_ms')::numeric))::numeric, 0) as p50_duration_ms,
ROUND((PERCENTILE_CONT(0.95) WITHIN GROUP (ORDER BY (tool->>'duration_ms')::numeric))::numeric, 0) as p95_duration_ms,
COUNT(*) FILTER (WHERE tool->>'is_error' = 'true') as error_count
FROM tool_calls
GROUP BY tool->>'name'
Expand All @@ -1006,7 +1006,7 @@ export class ThreadService {
`SELECT
t.channel,
COUNT(m.message_id) as message_count,
ROUND(AVG(m.latency_ms) FILTER (WHERE m.role = 'assistant')::numeric, 0) as avg_latency_ms
ROUND((AVG(m.latency_ms) FILTER (WHERE m.role = 'assistant'))::numeric, 0) as avg_latency_ms
FROM addie_threads t
JOIN addie_thread_messages m ON t.thread_id = m.thread_id
WHERE m.created_at > NOW() - make_interval(days => $1)
Expand All @@ -1025,7 +1025,7 @@ export class ThreadService {
`SELECT
DATE_TRUNC('day', created_at)::date::text as date,
COUNT(*) as message_count,
ROUND(AVG(latency_ms) FILTER (WHERE role = 'assistant')::numeric, 0) as avg_latency_ms,
ROUND((AVG(latency_ms) FILTER (WHERE role = 'assistant'))::numeric, 0) as avg_latency_ms,
COALESCE(SUM(tokens_input), 0) + COALESCE(SUM(tokens_output), 0) as total_tokens
FROM addie_thread_messages
WHERE created_at > NOW() - make_interval(days => $1)
Expand Down