v0.52 Improve handling and look at how we can improve ActiveRecord config for PostgreSQL and avoid insufficient pool size

This commit is contained in:
David Sainty 2024-08-19 21:33:37 +10:00
parent e37ba1c8a9
commit af3c46e1cf

View File

@ -1,7 +1,7 @@
# Federated Computer, Inc.
# David Sainty <saint@federated.computer> 2024 A.D.
# Gossamer Threads to Discourse -- Migration-Import Script
# v0.51 Fix mysql bug
# v0.52 Improve handling and look at how we can improve ActiveRecord config for PostgreSQL and avoid insufficient pool size.
require 'mysql2'
require 'open-uri'
@ -30,18 +30,6 @@ class GossamerForumsImporter < ImportScripts::Base
super
begin
# # Database configuration for ActiveRecord
# # This is not used, except for pool size... issue with our Bitnami Discourse?
# ActiveRecord::Base.establish_connection(
# adapter: 'postgresql',
# database: 'slowtwitch',
# username: 'admin',
# password: "yxnh93Ybbz2Nm8#mp28zCVv",
# host: 'slowtwitch.northend.network',
# pool: 20, # Adjust based on concurrency needs
# timeout: 5000
# )
# Initialize MySQL client to connect to Gossamer Forums database
@mysql_client = Mysql2::Client.new(
host: "slowtwitch.northend.network",
@ -1087,21 +1075,29 @@ class GossamerForumsImporter < ImportScripts::Base
# Get list of TOPICS / OP posts, i.e. post ids that have no parent / root id - SELECT post_id FROM gforum_Post WHERE post_root_id = 0;
def threaded_topic_import
# Define the custom connection pool settings
custom_pool = ActiveRecord::ConnectionAdapters::ConnectionPool.new(
ActiveRecord::Base.connection_pool.spec.to_h.merge(pool: 40, timeout: 5000)
)
# Register the custom connection pool under a unique identifier
ActiveRecord::Base.connection_handler.connection_pools['CustomPool'] = custom_pool
# Use CachedThreadPool for dynamic thread management
#### pool = Concurrent::CachedThreadPool.new
###### pool = Concurrent::FixedThreadPool.new(7)
pool = Concurrent::FixedThreadPool.new(7)
pool = Concurrent::FixedThreadPool.new(20)
# Define the connection pool inside the method
###### mariadb_pool = ConnectionPool.new(size: 14, timeout: 100) do
mariadb_pool = ConnectionPool.new(size: 40, timeout: 100) do
Mysql2::Client.new(
host: "slowtwitch.northend.network",
username: "admin",
password: "yxnh93Ybbz2Nm8#mp28zCVv",
database: "slowtwitch"
)
end
#### mariadb_pool = ConnectionPool.new(size: 40, timeout: 100) do
#### Mysql2::Client.new(
#### host: "slowtwitch.northend.network",
#### username: "admin",
#### password: "yxnh93Ybbz2Nm8#mp28zCVv",
#### database: "slowtwitch"
#### )
#### end
# The query selects post_ids from gforum_Post where post_root_id is 0, meaning these posts are the topic starters (OPs).
# Execute the query and fetch the result
@ -1184,7 +1180,8 @@ class GossamerForumsImporter < ImportScripts::Base
end
# Use connection pooling for PostgreSQL and synchronize access to shared resources
ActiveRecord::Base.connection_pool.with_connection do
# ActiveRecord::Base.connection_pool.with_connection do
ActiveRecord::Base.connected_to(pool: 'CustomPool') do
post_status = fetch_post_status(post_id)
if post_status.nil? || post_status == 0
puts "Starting import for post_id #{post_id}"
@ -1203,10 +1200,28 @@ class GossamerForumsImporter < ImportScripts::Base
sqlite_mutex.synchronize do
mark_post_as_failed(post_id)
end
if e.message =~ /MySQL client is not connected/ || e.message =~ /This connection is in use by/
sleep(1)
puts "Reconnecting to MySQL for post ID #{post_id} due to connection loss..."
retry
case e.message
when /MySQL client is not connected/, /This connection is in use by/
puts "Lost MySQL, retrying for post ID #{post_id}..."
# Add reconnection attempt again here... if it proves necessary?
retries ||= 0
retries += 1
if retries < 5
sleep(1)
retry
else
puts "Max retries reached for post ID #{post_id}"
end
when /could not obtain a connection from the pool/
puts "Connection pool exhausted, retrying for post ID #{post_id}..."
retries ||= 0
retries += 1
if retries < 5
sleep(1)
retry
else
puts "Max retries reached for post ID #{post_id}"
end
end
ensure
# Ensure the MariaDB connection is closed after processing
@ -1306,15 +1321,19 @@ class GossamerForumsImporter < ImportScripts::Base
# Check if the topic has already been imported using the custom field 'original_gossamer_id'
unless TopicCustomField.exists?(name: 'original_gossamer_id', value: row['post_id'])
puts "TIJ EE post_id #{post_id}"
ActiveRecord::Base.transaction do
# ActiveRecord::Base.transaction do
## ActiveRecord::Base.connected_to(pool: 'CustomPool') do
## ActiveRecord::Base.transaction do
# Create the new topic in Discourse
begin
suffix = 1
topic_created = false
ActiveRecord::Base.connected_to(pool: 'CustomPool') do
ActiveRecord::Base.transaction do
while !topic_created
begin
puts "TIJ FF post_id #{post_id}"
puts "TIJ FF post_id #{post_id}"
puts "CREATE TOPIC unique_title #{unique_title} title #{title} discourse_user_id #{discourse_user_id} category_id #{discourse_category_id}"
topic = Topic.create!(
title: unique_title,
@ -1340,6 +1359,8 @@ class GossamerForumsImporter < ImportScripts::Base
# puts e.backtrace.join("\n") # Print the full stack trace
end
end
end
end
# Workaround... take a copy of topic.id
current_topic_id = topic.id
@ -1365,8 +1386,10 @@ class GossamerForumsImporter < ImportScripts::Base
update_db_topic_post_numbers(current_topic_id, post_number)
end
puts "TIJ GG post_id #{post_id}"
puts "TIJ GG post_id #{post_id}"
ActiveRecord::Base.connected_to(pool: 'CustomPool') do
ActiveRecord::Base.transaction do
# Create the initial post in the new topic
post = Post.create!(
topic_id: current_topic_id,
@ -1379,6 +1402,8 @@ class GossamerForumsImporter < ImportScripts::Base
)
post.custom_fields['original_gossamer_id'] = row['post_id']
post.save!
end
end
sqlite_mutex.synchronize do
# Increment the post count for the topic and user
@ -1386,9 +1411,14 @@ class GossamerForumsImporter < ImportScripts::Base
update_db_user_post_count(discourse_user_id, fetch_db_user_post_count(discourse_user_id).to_i + 1)
end
puts "TIJ HH post_id #{post_id}"
puts "TIJ HH post_id #{post_id}"
ActiveRecord::Base.connected_to(pool: 'CustomPool') do
ActiveRecord::Base.transaction do
# Handle any attachments associated with the post
handle_post_attachments(row['post_id'], post, discourse_user_id, mysql_client)
end
end
# Create URL mappings for the new topic
new_url = "https://new/t/#{topic.slug}/#{current_topic_id}"
@ -1412,7 +1442,7 @@ class GossamerForumsImporter < ImportScripts::Base
puts "reply_user_id is NOW Former_User id #{reply_user_id} for reply post_id #{reply_row['post_id']}"
end
puts "TIJ II post_id #{post_id}"
puts "TIJ II post_id #{post_id}"
# Sanitize and prepare the reply message for Discourse
sanitized_reply_message = sanitize_post_message(reply_row['post_message'])
@ -1437,6 +1467,8 @@ class GossamerForumsImporter < ImportScripts::Base
end
puts "TIJ JJ post_id #{post_id} reply post_id #{reply_row['post_id']} reply_post_views #{reply_post_views || 0} post_number #{post_number} current_topic_id #{current_topic_id} reply_post_views #{reply_post_views || 0}"
ActiveRecord::Base.connected_to(pool: 'CustomPool') do
ActiveRecord::Base.transaction do
# Create the reply post in the existing topic
post = Post.create!(
topic_id: current_topic_id,
@ -1449,8 +1481,10 @@ class GossamerForumsImporter < ImportScripts::Base
)
post.custom_fields['original_gossamer_id'] = reply_row['post_id']
post.save!
puts "TIJ KK post_id #{post_id}"
end
end
puts "TIJ KK post_id #{post_id}"
# Increment the post count for the topic and user
sqlite_mutex.synchronize do
update_db_topic_post_count(current_topic_id, fetch_db_topic_post_count(current_topic_id).to_i + 1)
@ -1465,8 +1499,12 @@ class GossamerForumsImporter < ImportScripts::Base
end
end
ActiveRecord::Base.connected_to(pool: 'CustomPool') do
ActiveRecord::Base.transaction do
# Handle any attachments associated with the reply
handle_post_attachments(reply_row['post_id'], post, reply_user_id, mysql_client)
end
end
# # Update the highest processed post_id in the database (thread-safe)
# update_highest_processed_post_id_thread_safe(reply_row['post_id'])
@ -1485,7 +1523,8 @@ class GossamerForumsImporter < ImportScripts::Base
puts "Error importing topic with post_id #{row['post_id']}: #{e.message}"
raise ActiveRecord::Rollback
end
end
# end
# end
else
puts "Topic for post_id #{row['post_id']} already exists, skipping creation."
end