perf: shorten Intercom import jobs (#15051)

## Description

Reduces sustained load during large Intercom imports by limiting
conversation list pages to 10 while retaining 50-contact pages. Cursor
and provider total-count behavior remain unchanged.

The one-minute progress heartbeat now lives in parent PR #15050 because
stalled-import detection must be safe when that PR is deployed
independently. This child PR therefore contains only the smaller-page
delta.

This is Phase 1, Task 2 of the [Intercom import optimization plan
(CW-7615)](https://linear.app/chatwoot/issue/CW-7615/optimize-intercom-import-reliability-and-bulk-message-ingestion).

This PR is stacked on #15050 and should be reviewed as the two-file
delta from `codex/cw-7519-intercom-stalled-retry-15m`.

## Closes

-
[CW-7615](https://linear.app/chatwoot/issue/CW-7615/optimize-intercom-import-reliability-and-bulk-message-ingestion)

## Type of change

- [x] Bug fix (non-breaking change which fixes an issue)

## How to test

1. Start an Intercom import containing contacts and conversations across
multiple source pages.
2. Confirm contact list requests retain a page size of 50.
3. Confirm conversation list requests use a page size of 10.
4. Confirm page cursors and displayed provider totals continue advancing
normally.
5. Confirm the parent PR heartbeat keeps long conversations fresh while
these shorter pages are processed.

## Checklist

- [x] My code follows the style guidelines of this project
- [x] I have performed a self-review of my code
- [x] I have added tests that prove the change is effective
- [x] New and existing focused tests pass locally with my changes
This commit is contained in:
Sony Mathew
2026-07-23 23:38:01 +05:30
committed by GitHub
parent ea783a89ea
commit b6efdae243
4 changed files with 777 additions and 123 deletions
@@ -66,12 +66,12 @@ RSpec.describe DataImports::Intercom::Importer do
before do
account.enable_features!('data_import')
allow(DataImports::Intercom::Client).to receive(:new).with(access_token: 'intercom-token').and_return(client)
allow(client).to receive(:list_contacts).with(starting_after: nil).and_return(
allow(client).to receive(:list_contacts).with(starting_after: nil, per_page: 50).and_return(
'data' => [contact_payload],
'total_count' => 1,
'pages' => { 'next' => nil }
)
allow(client).to receive(:list_conversations).with(starting_after: nil).and_return(
allow(client).to receive(:list_conversations).with(starting_after: nil, per_page: 10).and_return(
'conversations' => [{ 'id' => 'conversation_1' }],
'total_count' => 1,
'pages' => { 'next' => nil }
@@ -188,14 +188,49 @@ RSpec.describe DataImports::Intercom::Importer do
expect(item.metadata['message_total_contribution']).to eq(3)
end
it 'uses smaller conversation pages while retaining the contact page size' do
importer = described_class.new(data_import: data_import)
importer.import_contacts_page
importer.import_conversations_page
expect(client).to have_received(:list_contacts).with(starting_after: nil, per_page: 50)
expect(client).to have_received(:list_conversations).with(starting_after: nil, per_page: 10)
end
it 'reconciles imported message stats from same-run mappings on retry' do
described_class.new(data_import: data_import).import_conversations_page
stats = data_import.reload.stats.deep_dup
stats['messages']['imported'] = 0
data_import.update!(stats: stats)
importer = described_class.new(data_import: data_import)
expect(importer).to receive(:reconcile_message_stats).once.ordered.and_call_original
expect(importer).to receive(:update_cursor).with('conversations', nil).once.ordered.and_call_original
importer.import_conversations_page
expect(data_import.reload.stats.dig('messages', 'imported')).to eq(3)
end
it 'retries a query timeout during deferred message stats reconciliation', :aggregate_failures do
described_class.new(data_import: data_import).import_conversations_page
stats = data_import.reload.stats.deep_dup
stats['messages']['imported'] = 0
data_import.update!(stats: stats)
importer = described_class.new(data_import: data_import)
reconciliation_attempts = 0
allow(importer).to receive(:sleep)
allow(importer).to receive(:reconcile_message_stats).and_wrap_original do |method|
reconciliation_attempts += 1
raise ActiveRecord::QueryCanceled, 'statement timeout' if reconciliation_attempts == 1
method.call
end
importer.import_conversations_page
expect(reconciliation_attempts).to eq(2)
expect(importer).to have_received(:sleep).with(be_between(0.2, 0.5)).once
expect(data_import.reload.stats.dig('messages', 'imported')).to eq(3)
end
@@ -267,7 +302,7 @@ RSpec.describe DataImports::Intercom::Importer do
it 'stops an in-flight page when a newer import run takes over', :aggregate_failures do
run_id = 'intercom-run-1'
data_import.update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => run_id })
allow(client).to receive(:list_conversations).with(starting_after: nil).and_return(
allow(client).to receive(:list_conversations).with(starting_after: nil, per_page: 10).and_return(
'conversations' => [{ 'id' => 'conversation_1' }, { 'id' => 'conversation_2' }],
'pages' => { 'next' => { 'starting_after' => 'next-conversation-cursor' } }
)
@@ -326,6 +361,49 @@ RSpec.describe DataImports::Intercom::Importer do
end
end
it 'does not persist stale stats when a newer run takes over during the final part', :aggregate_failures do
run_id = 'intercom-run-1'
data_import.update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => run_id })
importer = described_class.new(data_import: data_import, run_id: run_id)
allow(importer).to receive(:create_message).and_wrap_original do |method, *args|
method.call(*args).tap do
entry = args[2]
next unless entry.part['id'] == 'part_2'
DataImport.find(data_import.id).update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => 'new-run' })
end
end
importer.import_conversations_page
conversation = account.conversations.find_by!(identifier: 'intercom:conversation_1')
expect(conversation.messages.count).to eq(3)
expect(data_import.reload.stats.dig('conversations', 'imported')).to eq(0)
end
it 'reconciles conversation stats when a superseded run is retried', :aggregate_failures do
freeze_time do
run_id = 'intercom-run-1'
next_run_id = 'intercom-run-2'
data_import.update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => run_id })
importer = described_class.new(data_import: data_import, run_id: run_id)
allow(importer).to receive(:create_message) do
data_import.update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => next_run_id })
travel 1.minute
end
importer.import_conversations_page
expect(data_import.reload.stats.dig('conversations', 'imported')).to eq(0)
retry_importer = described_class.new(data_import: data_import, run_id: next_run_id)
retry_importer.import_conversations_page
retry_importer.finish!
expect(data_import.reload.stats.dig('conversations', 'imported')).to eq(1)
expect(data_import.processed_records).to eq(5)
end
end
it 'rolls back a newly inserted conversation when mapping persistence fails', :aggregate_failures do
importer = described_class.new(data_import: data_import)
allow(importer).to receive(:record_mapping).and_wrap_original do |method, object_type, source_id, record, metadata:|
@@ -363,11 +441,7 @@ RSpec.describe DataImports::Intercom::Importer do
it 'rolls back a newly inserted message when mapping persistence fails', :aggregate_failures do
importer = described_class.new(data_import: data_import)
allow(importer).to receive(:record_mapping).and_wrap_original do |method, object_type, source_id, record, metadata:|
raise StandardError, 'mapping failed' if object_type == 'message'
method.call(object_type, source_id, record, metadata: metadata)
end
allow(importer).to receive(:record_message_mapping).and_raise(StandardError, 'mapping failed')
importer.import_conversations_page
@@ -379,6 +453,79 @@ RSpec.describe DataImports::Intercom::Importer do
)
expect(error).to have_attributes(error_code: 'StandardError', message: 'mapping failed')
end
it 'retries a contact query timeout after rolling back the first transaction', :aggregate_failures do
importer = described_class.new(data_import: data_import)
mapping_attempts = 0
allow(importer).to receive(:sleep)
allow(importer).to receive(:record_mapping).and_wrap_original do |method, object_type, source_id, record, metadata:|
if object_type == 'contact'
mapping_attempts += 1
raise ActiveRecord::QueryCanceled, 'statement timeout' if mapping_attempts == 1
end
method.call(object_type, source_id, record, metadata: metadata)
end
importer.import_contacts_page
expect(mapping_attempts).to eq(2)
expect(importer).to have_received(:sleep).with(be_between(0.2, 0.5)).once
expect(account.contacts.where(email: 'customer@example.com').count).to eq(1)
expect(data_import.mappings.where(source_object_type: 'contact', source_object_id: 'contact_1').count).to eq(1)
expect(data_import.import_errors).to be_empty
expect(data_import.reload.stats.dig('contacts', 'imported')).to eq(1)
end
it 'retries a message query timeout after rolling back the first transaction', :aggregate_failures do
importer = described_class.new(data_import: data_import)
mapping_attempts = 0
target_source_id = 'conversation:conversation_1:part:part_1'
allow(importer).to receive(:sleep)
allow(importer).to receive(:record_message_mapping).and_wrap_original do |method, entry, message|
if entry.source_id == target_source_id
mapping_attempts += 1
raise ActiveRecord::QueryCanceled, 'statement timeout' if mapping_attempts == 1
end
method.call(entry, message)
end
importer.import_conversations_page
conversation = account.conversations.find_by!(identifier: 'intercom:conversation_1')
expect(mapping_attempts).to eq(2)
expect(importer).to have_received(:sleep).with(be_between(0.2, 0.5)).once
expect(conversation.messages.where(source_id: "intercom:#{target_source_id}").count).to eq(1)
expect(data_import.mappings.where(source_object_type: 'message', source_object_id: target_source_id).count).to eq(1)
expect(data_import.import_errors).to be_empty
expect(data_import.reload.stats.dig('messages', 'imported')).to eq(3)
end
it 'records one message failure only after the query timeout retry is exhausted', :aggregate_failures do
importer = described_class.new(data_import: data_import)
mapping_attempts = 0
target_source_id = 'conversation:conversation_1:part:part_1'
allow(importer).to receive(:sleep)
allow(importer).to receive(:record_message_mapping).and_wrap_original do |method, entry, message|
if entry.source_id == target_source_id
mapping_attempts += 1
raise ActiveRecord::QueryCanceled, 'statement timeout'
end
method.call(entry, message)
end
importer.import_conversations_page
conversation = account.conversations.find_by!(identifier: 'intercom:conversation_1')
error = data_import.import_errors.find_by!(source_object_type: 'message', source_object_id: target_source_id)
expect(mapping_attempts).to eq(2)
expect(importer).to have_received(:sleep).with(be_between(0.2, 0.5)).once
expect(conversation.messages.where(source_id: "intercom:#{target_source_id}")).to be_empty
expect(error).to have_attributes(error_code: 'ActiveRecord::QueryCanceled', message: 'statement timeout')
expect(data_import.reload.stats.dig('errors', 'count')).to eq(1)
end
end
describe '#finish!' do
@@ -467,6 +614,57 @@ RSpec.describe DataImports::Intercom::Importer do
expect(next_data_import.import_errors.skip_logs.pluck(:details).map { |details| details['reason'] }.uniq).to eq(['already_imported'])
end
it 'reconciles skipped stats when a superseded run is retried', :aggregate_failures do
described_class.new(data_import: data_import).perform
run_id = 'intercom-run-1'
next_run_id = 'intercom-run-2'
next_data_import.update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => run_id })
importer = described_class.new(data_import: next_data_import, run_id: run_id)
allow(importer).to receive(:skip_existing_message_mapping).and_wrap_original do |method, *args|
method.call(*args)
part = args[2]
next unless part['id'] == 'part_2'
DataImport.find(next_data_import.id).update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => next_run_id })
end
importer.import_conversations_page
expect(next_data_import.reload.stats.dig('messages', 'skipped')).to eq(0)
retry_importer = described_class.new(data_import: next_data_import, run_id: next_run_id)
retry_importer.import_conversations_page
retry_importer.finish!
expect(next_data_import.reload.stats.dig('conversations', 'skipped')).to eq(1)
expect(next_data_import.stats.dig('messages', 'skipped')).to eq(3)
expect(next_data_import.total_records).to eq(5)
end
it 'keeps skipped contact stats when a query timeout is retried after the item update', :aggregate_failures do
described_class.new(data_import: data_import).perform
importer = described_class.new(data_import: next_data_import)
contact_log_attempts = 0
allow(importer).to receive(:sleep)
allow(importer).to receive(:record_already_imported_log).and_wrap_original do |method, **attributes|
if attributes[:source_object_type] == 'contact'
contact_log_attempts += 1
raise ActiveRecord::QueryCanceled, 'statement timeout' if contact_log_attempts == 1
end
method.call(**attributes)
end
importer.import_contacts_page
contact_item = next_data_import.items.find_by!(source_object_type: 'contact', source_object_id: 'contact_1')
expect(contact_log_attempts).to eq(2)
expect(importer).to have_received(:sleep).with(be_between(0.2, 0.5)).once
expect(contact_item).to be_skipped
expect(next_data_import.reload.stats.dig('contacts', 'skipped')).to eq(1)
expect(next_data_import.import_errors.skip_logs.exists?(data_import_item: contact_item)).to be(true)
end
it 'recreates messages when existing message mappings point to deleted records', :aggregate_failures do
described_class.new(data_import: data_import).perform
conversation = account.conversations.find_by!(identifier: 'intercom:conversation_1')
@@ -492,6 +690,33 @@ RSpec.describe DataImports::Intercom::Importer do
expect(message_mappings.filter_map(&:chatwoot_record).count).to eq(3)
end
it 'repairs a missing mapping without recreating the existing message', :aggregate_failures do
described_class.new(data_import: data_import).import_conversations_page
conversation = account.conversations.find_by!(identifier: 'intercom:conversation_1')
message = conversation.messages.find_by!(source_id: 'intercom:conversation:conversation_1:part:part_1')
DataImportMapping.find_by!(
account: account,
source_provider: 'intercom',
source_object_type: 'message',
source_object_id: 'conversation:conversation_1:part:part_1'
).destroy!
importer = described_class.new(data_import: data_import)
allow(importer).to receive(:find_mapping).and_call_original
importer.import_conversations_page
repaired_mapping = DataImportMapping.find_by!(
account: account,
source_provider: 'intercom',
source_object_type: 'message',
source_object_id: 'conversation:conversation_1:part:part_1'
)
expect(conversation.messages.where(source_id: message.source_id).count).to eq(1)
expect(repaired_mapping.chatwoot_record).to eq(message)
expect(importer).not_to have_received(:find_mapping).with('message', anything)
expect(data_import.reload.stats.dig('messages', 'imported')).to eq(3)
end
it 'updates conversation activity when a later import adds new messages to the mapped conversation', :aggregate_failures do
new_part = {
'id' => 'part_3',
@@ -554,7 +779,11 @@ RSpec.describe DataImports::Intercom::Importer do
end
it 'repairs the item and imported count on retry', :aggregate_failures do
described_class.new(data_import: data_import).import_contacts_page
importer = described_class.new(data_import: data_import)
expect(importer).to receive(:reconcile_item_stats).with('contact').once.ordered.and_call_original
expect(importer).to receive(:update_cursor).with('contacts', nil).once.ordered.and_call_original
importer.import_contacts_page
item = data_import.items.find_by!(source_object_type: 'contact', source_object_id: 'contact_1')
expect(item).to be_imported
@@ -937,6 +1166,32 @@ RSpec.describe DataImports::Intercom::Importer do
expect(data_import.reload).to be_completed_with_errors
expect(data_import.stats.dig('errors', 'count')).to eq(1)
end
it 'reconciles error stats when a superseded run is retried', :aggregate_failures do
run_id = 'intercom-run-1'
next_run_id = 'intercom-run-2'
data_import.update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => run_id })
importer = described_class.new(data_import: data_import, run_id: run_id)
allow(importer).to receive(:import_message).and_wrap_original do |method, *args|
method.call(*args).tap do
entry = args[2]
next unless entry.part['id'] == 'part_2'
DataImport.find(data_import.id).update!(source_metadata: { DataImport::ACTIVE_INTERCOM_IMPORT_RUN_ID_KEY => next_run_id })
end
end
importer.import_conversations_page
expect(data_import.reload.stats.dig('errors', 'count')).to eq(0)
retry_importer = described_class.new(data_import: data_import, run_id: next_run_id)
retry_importer.import_conversations_page
retry_importer.finish!
expect(data_import.reload).to be_completed_with_errors
expect(data_import.stats.dig('errors', 'count')).to eq(1)
expect(data_import.total_records).to eq(6)
end
end
context 'when the conversation parts total matches the returned parts' do
@@ -983,6 +1238,8 @@ RSpec.describe DataImports::Intercom::Importer do
end
context 'when a specific Intercom message part fails to persist' do
let(:insert_attempts) { [] }
let(:conversation_payload) do
super().deep_merge(
'conversation_parts' => {
@@ -1003,7 +1260,10 @@ RSpec.describe DataImports::Intercom::Importer do
before do
allow(Message).to receive(:insert_all!).and_wrap_original do |method, records, **kwargs|
raise ActiveRecord::StatementInvalid, 'bad message' if records.first[:source_id] == 'intercom:conversation:conversation_1:part:bad_part'
if records.first[:source_id] == 'intercom:conversation:conversation_1:part:bad_part'
insert_attempts << records.first[:source_id]
raise ActiveRecord::StatementInvalid, 'bad message'
end
method.call(records, **kwargs)
end
@@ -1024,6 +1284,7 @@ RSpec.describe DataImports::Intercom::Importer do
)
expect(data_import.reload).to be_completed_with_errors
expect(data_import.stats.dig('errors', 'count')).to eq(1)
expect(insert_attempts.one?).to be(true)
end
end
end
@@ -0,0 +1,178 @@
require 'rails_helper'
RSpec.describe DataImports::Intercom::MessageBatchBuilder do
let(:account) { create(:account) }
let(:data_import) { create(:data_import, :intercom, account: account) }
let(:conversation) { create(:conversation, account: account) }
let(:source_conversation) do
{
'id' => 'conversation_1',
'created_at' => 1_700_000_000,
'source' => {
'id' => 'source_1',
'part_type' => 'conversation',
'body' => '<p>Initial message</p>',
'created_at' => 1_700_000_000
},
'conversation_parts' => {
'conversation_parts' => [
{
'id' => 'part_1',
'part_type' => 'comment',
'body' => '<p>First reply</p>',
'created_at' => 1_700_000_000
},
{
'id' => 'part_2',
'part_type' => 'note',
'body' => '<p>Internal note</p>',
'created_at' => 1_700_000_100
}
]
}
}
end
let(:builder) do
described_class.new(
data_import: data_import,
conversation: conversation,
source_conversation: source_conversation
)
end
it 'preserves source order and prefetches mappings and messages once per batch', :aggregate_failures do
sql_queries = []
subscriber = lambda do |_name, _start, _finish, _id, payload|
sql_queries << payload unless payload[:name] == 'SCHEMA'
end
batch_builder = builder
batch = ActiveSupport::Notifications.subscribed(subscriber, 'sql.active_record') { batch_builder.perform }
expect(batch.entries.map(&:source_id)).to eq(
%w[
conversation:conversation_1:source:source_1
conversation:conversation_1:part:part_1
conversation:conversation_1:part:part_2
]
)
expect(batch.entries.map(&:position)).to eq([0, 1, 2])
expect(batch.entries.map(&:classification)).to all(eq(:new_message))
expect(sql_queries.count { |query| query[:name] == 'DataImportMapping Load' }).to eq(1)
message_queries = sql_queries.select { |query| query[:name] == 'Message Load' }
expect(message_queries.size).to eq(1), message_queries.pluck(:sql).join("\n")
end
it 'returns an empty batch without prefetch queries when the conversation has no messages' do
source_conversation['source'] = nil
source_conversation['conversation_parts']['conversation_parts'] = []
expect(DataImportMapping).not_to receive(:where)
expect(Message).not_to receive(:where)
expect(builder.perform.entries).to be_empty
end
it 'classifies a live mapping from the current import as already handled' do
message = create(
:message,
account: account,
conversation: conversation,
inbox: conversation.inbox,
source_id: 'intercom:conversation:conversation_1:part:part_1'
)
DataImportMapping.create!(
account: account,
data_import: data_import,
source_provider: 'intercom',
source_object_type: 'message',
source_object_id: 'conversation:conversation_1:part:part_1',
chatwoot_record_type: 'Message',
chatwoot_record_id: message.id,
metadata: {}
)
entry = builder.perform.entries.find { |batch_entry| batch_entry.source_id.end_with?('part:part_1') }
expect(entry).to have_attributes(classification: :current_import, message: message)
end
it 'classifies a live mapping from a previous import without changing its owner' do
previous_import = create(:data_import, :intercom, account: account)
message = create(
:message,
account: account,
conversation: conversation,
inbox: conversation.inbox,
source_id: 'intercom:conversation:conversation_1:part:part_1'
)
mapping = DataImportMapping.create!(
account: account,
data_import: previous_import,
source_provider: 'intercom',
source_object_type: 'message',
source_object_id: 'conversation:conversation_1:part:part_1',
chatwoot_record_type: 'Message',
chatwoot_record_id: message.id,
metadata: {}
)
entry = builder.perform.entries.find { |batch_entry| batch_entry.source_id.end_with?('part:part_1') }
expect(entry).to have_attributes(classification: :previous_import, mapping: mapping, message: message)
expect(mapping.reload.data_import).to eq(previous_import)
end
it 'classifies a mapping whose message was deleted as repairable' do
mapping = DataImportMapping.create!(
account: account,
data_import: data_import,
source_provider: 'intercom',
source_object_type: 'message',
source_object_id: 'conversation:conversation_1:part:part_1',
chatwoot_record_type: 'Message',
chatwoot_record_id: 0,
metadata: {}
)
entry = builder.perform.entries.find { |batch_entry| batch_entry.source_id.end_with?('part:part_1') }
expect(entry).to have_attributes(classification: :repairable_stale_mapping, mapping: mapping, message: nil)
end
it 'classifies an existing conversation message without a mapping for repair' do
message = create(
:message,
account: account,
conversation: conversation,
inbox: conversation.inbox,
source_id: 'intercom:conversation:conversation_1:part:part_1'
)
entry = builder.perform.entries.find { |batch_entry| batch_entry.source_id.end_with?('part:part_1') }
expect(entry).to have_attributes(classification: :existing_message, mapping: nil, message: message)
end
it 'repairs a skipped mapping when the source part is now an activity' do
source_conversation.dig('conversation_parts', 'conversation_parts').first.merge!(
'part_type' => 'assignment',
'body' => nil,
'assigned_to' => { 'name' => 'Support' }
)
mapping = DataImportMapping.create!(
account: account,
data_import: data_import,
source_provider: 'intercom',
source_object_type: 'message',
source_object_id: 'conversation:conversation_1:part:part_1',
chatwoot_record_type: 'Conversation',
chatwoot_record_id: conversation.id,
metadata: { skipped: true }
)
entry = builder.perform.entries.find { |batch_entry| batch_entry.source_id.end_with?('part:part_1') }
expect(entry).to have_attributes(classification: :repairable_stale_mapping, mapping: mapping, message: nil)
end
end