frappe-ops-performance

Name: frappe-ops-performance
Author: Impertio-Studio

151stars49forks

How to install frappe-ops-performance?

impertio-studio/frappe_claude_skill_package/frappe-ops-performance

$npx -y skills add Impertio-Studio/Frappe_Claude_Skill_Package --skill frappe-ops-performance

Installs into the current project.

Summary

Use when tuning MariaDB, configuring Redis memory, sizing Gunicorn workers, setting up CDN, or profiling slow queries. Prevents performance bottlenecks from default configurations, memory exhaustion, and unoptimized database queries. Covers MariaDB tuning, Redis configuration, Gu

Files · 1

View on GitHub

SKILL.md

1	# Performance Tuning
2
3	Frappe/ERPNext performance depends on four layers: database (MariaDB), cache (Redis), application server (Gunicorn), and background workers (RQ). ALWAYS tune all four layers together — optimizing one while ignoring others creates new bottlenecks.
4
5	## Quick Reference
6
7	```bash
8	# Check system health
9	bench doctor
10
11	# Show pending background jobs
12	bench --site mysite.com show-pending-jobs
13
14	# Clear all caches
15	bench --site mysite.com clear-cache
16	bench --site mysite.com clear-website-cache
17
18	# Purge stuck background jobs
19	bench purge-jobs
20
21	# Enable MariaDB slow query log
22	# In /etc/mysql/mariadb.conf.d/50-server.cnf:
23	# slow_query_log = 1
24	# slow_query_log_file = /var/log/mysql/slow.log
25	# long_query_time = 1
26
27	# Check Gunicorn worker count
28	# In Procfile or supervisor config: -w [workers]
29	# Formula: workers = (2 * CPU_CORES) + 1
30	```
31
32	---
33
34	## Performance Decision Tree
35
36	```
37	What is slow?
38	\|
39	+-- Page loads are slow?
40	\| +-- Check Gunicorn workers (are they saturated?)
41	\| +-- Check MariaDB slow query log
42	\| +-- Check Redis memory (is cache evicting?)
43	\| +-- Enable CDN for static assets
44	\|
45	+-- Background jobs are delayed?
46	\| +-- bench doctor (check worker count and pending jobs)
47	\| +-- Increase RQ worker count
48	\| +-- Check for long-running jobs blocking queues
49	\|
50	+-- Database queries are slow?
51	\| +-- Enable slow query log
52	\| +-- Run EXPLAIN on slow queries
53	\| +-- Add indexes on frequently filtered columns
54	\| +-- Use get_cached_value instead of get_value
55	\|
56	+-- Server runs out of memory?
57	\| +-- Reduce Gunicorn workers
58	\| +-- Set Redis maxmemory
59	\| +-- Check MariaDB innodb_buffer_pool_size
60	\| +-- Look for memory leaks in custom code
61	\|
62	+-- High CPU usage?
63	\| +-- Profile Python code (cProfile)
64	\| +-- Check for N+1 query patterns
65	\| +-- Review custom scheduled jobs
66	```
67
68	---
69
70	## MariaDB Tuning
71
72	### Critical Settings
73
74	```ini
75	# /etc/mysql/mariadb.conf.d/50-server.cnf
76	[mysqld]
77	# InnoDB buffer pool — MOST important setting
78	# Set to 50-70% of available RAM on dedicated DB server
79	# Set to 25-40% of RAM on shared server
80	innodb_buffer_pool_size = 2G
81
82	# Buffer pool instances (1 per GB of buffer pool)
83	innodb_buffer_pool_instances = 2
84
85	# Log file size (larger = better write performance, slower recovery)
86	innodb_log_file_size = 256M
87
88	# Flush method — use O_DIRECT to avoid double buffering
89	innodb_flush_method = O_DIRECT
90
91	# Character set (ALWAYS use utf8mb4 for Frappe)
92	character-set-server = utf8mb4
93	collation-server = utf8mb4_unicode_ci
94
95	# Key buffer for MyISAM (Frappe uses InnoDB, keep small)
96	key_buffer_size = 32M
97
98	# Query cache (DISABLE for MariaDB 10.4+ / MySQL 8.0+)
99	query_cache_type = 0
100	query_cache_size = 0
101
102	# Connection limits
103	max_connections = 200
104	wait_timeout = 600
105	interactive_timeout = 600
106
107	# Temp tables
108	tmp_table_size = 64M
109	max_heap_table_size = 64M
110
111	# Slow query log
112	slow_query_log = 1
113	slow_query_log_file = /var/log/mysql/slow.log
114	long_query_time = 1
115	```
116
117	### Slow Query Analysis
118
119	```bash
120	# Enable slow query log (runtime, no restart needed)
121	SET GLOBAL slow_query_log = 1;
122	SET GLOBAL long_query_time = 1;
123
124	# Analyze slow queries with mysqldumpslow
125	mysqldumpslow -t 10 -s c /var/log/mysql/slow.log
126	# -t 10: top 10 queries
127	# -s c: sort by count (use -s t for total time)
128
129	# Use EXPLAIN to analyze specific queries
130	EXPLAIN SELECT * FROM `tabSales Invoice` WHERE customer = 'ABC';
131	# Look for: type=ALL (full table scan), rows > 10000, Using filesort
132	```
133
134	### Index Optimization
135
136	```sql
137	-- Check for missing indexes on frequently filtered columns
138	SHOW INDEX FROM `tabSales Invoice`;
139
140	-- Add index for common filter patterns
141	ALTER TABLE `tabSales Invoice` ADD INDEX idx_customer_date (customer, posting_date);
142
143	-- Frappe way: add index via DocType definition
144	-- In doctype JSON: set "in_list_view" or "search_index" on fields
145	-- OR use hooks.py:
146	-- after_migrate = ["myapp.patches.add_custom_indexes"]
147	```
148
149	---
150
151	## Redis Configuration
152
153	### Memory Management
154
155	```conf
156	# /etc/redis/redis.conf (or bench config/redis_cache.conf)
157
158	# Set maximum memory — NEVER let Redis use all available RAM
159	maxmemory 512mb
160
161	# Eviction policy — allkeys-lru is best for cache use
162	maxmemory-policy allkeys-lru
163
164	# Disable persistence for cache Redis (performance boost)
165	save ""
166	appendonly no
167	```
168
169	### Fr

People who liked this also viewed

6 picks

Type

frappe-ops-performance

151stars49forks

View on GitHub

How to install frappe-ops-performance?

impertio-studio/frappe_claude_skill_package/frappe-ops-performance

$npx -y skills add Impertio-Studio/Frappe_Claude_Skill_Package --skill frappe-ops-performance

Installs into the current project.

Summary

Files · 1

View on GitHub

SKILL.md

1	# Performance Tuning
2
3	Frappe/ERPNext performance depends on four layers: database (MariaDB), cache (Redis), application server (Gunicorn), and background workers (RQ). ALWAYS tune all four layers together — optimizing one while ignoring others creates new bottlenecks.
4
5	## Quick Reference
6
7	```bash
8	# Check system health
9	bench doctor
10
11	# Show pending background jobs
12	bench --site mysite.com show-pending-jobs
13
14	# Clear all caches
15	bench --site mysite.com clear-cache
16	bench --site mysite.com clear-website-cache
17
18	# Purge stuck background jobs
19	bench purge-jobs
20
21	# Enable MariaDB slow query log
22	# In /etc/mysql/mariadb.conf.d/50-server.cnf:
23	# slow_query_log = 1
24	# slow_query_log_file = /var/log/mysql/slow.log
25	# long_query_time = 1
26
27	# Check Gunicorn worker count
28	# In Procfile or supervisor config: -w [workers]
29	# Formula: workers = (2 * CPU_CORES) + 1
30	```
31
32	---
33
34	## Performance Decision Tree
35
36	```
37	What is slow?
38	\|
39	+-- Page loads are slow?
40	\| +-- Check Gunicorn workers (are they saturated?)
41	\| +-- Check MariaDB slow query log
42	\| +-- Check Redis memory (is cache evicting?)
43	\| +-- Enable CDN for static assets
44	\|
45	+-- Background jobs are delayed?
46	\| +-- bench doctor (check worker count and pending jobs)
47	\| +-- Increase RQ worker count
48	\| +-- Check for long-running jobs blocking queues
49	\|
50	+-- Database queries are slow?
51	\| +-- Enable slow query log
52	\| +-- Run EXPLAIN on slow queries
53	\| +-- Add indexes on frequently filtered columns
54	\| +-- Use get_cached_value instead of get_value
55	\|
56	+-- Server runs out of memory?
57	\| +-- Reduce Gunicorn workers
58	\| +-- Set Redis maxmemory
59	\| +-- Check MariaDB innodb_buffer_pool_size
60	\| +-- Look for memory leaks in custom code
61	\|
62	+-- High CPU usage?
63	\| +-- Profile Python code (cProfile)
64	\| +-- Check for N+1 query patterns
65	\| +-- Review custom scheduled jobs
66	```
67
68	---
69
70	## MariaDB Tuning
71
72	### Critical Settings
73
74	```ini
75	# /etc/mysql/mariadb.conf.d/50-server.cnf
76	[mysqld]
77	# InnoDB buffer pool — MOST important setting
78	# Set to 50-70% of available RAM on dedicated DB server
79	# Set to 25-40% of RAM on shared server
80	innodb_buffer_pool_size = 2G
81
82	# Buffer pool instances (1 per GB of buffer pool)
83	innodb_buffer_pool_instances = 2
84
85	# Log file size (larger = better write performance, slower recovery)
86	innodb_log_file_size = 256M
87
88	# Flush method — use O_DIRECT to avoid double buffering
89	innodb_flush_method = O_DIRECT
90
91	# Character set (ALWAYS use utf8mb4 for Frappe)
92	character-set-server = utf8mb4
93	collation-server = utf8mb4_unicode_ci
94
95	# Key buffer for MyISAM (Frappe uses InnoDB, keep small)
96	key_buffer_size = 32M
97
98	# Query cache (DISABLE for MariaDB 10.4+ / MySQL 8.0+)
99	query_cache_type = 0
100	query_cache_size = 0
101
102	# Connection limits
103	max_connections = 200
104	wait_timeout = 600
105	interactive_timeout = 600
106
107	# Temp tables
108	tmp_table_size = 64M
109	max_heap_table_size = 64M
110
111	# Slow query log
112	slow_query_log = 1
113	slow_query_log_file = /var/log/mysql/slow.log
114	long_query_time = 1
115	```
116
117	### Slow Query Analysis
118
119	```bash
120	# Enable slow query log (runtime, no restart needed)
121	SET GLOBAL slow_query_log = 1;
122	SET GLOBAL long_query_time = 1;
123
124	# Analyze slow queries with mysqldumpslow
125	mysqldumpslow -t 10 -s c /var/log/mysql/slow.log
126	# -t 10: top 10 queries
127	# -s c: sort by count (use -s t for total time)
128
129	# Use EXPLAIN to analyze specific queries
130	EXPLAIN SELECT * FROM `tabSales Invoice` WHERE customer = 'ABC';
131	# Look for: type=ALL (full table scan), rows > 10000, Using filesort
132	```
133
134	### Index Optimization
135
136	```sql
137	-- Check for missing indexes on frequently filtered columns
138	SHOW INDEX FROM `tabSales Invoice`;
139
140	-- Add index for common filter patterns
141	ALTER TABLE `tabSales Invoice` ADD INDEX idx_customer_date (customer, posting_date);
142
143	-- Frappe way: add index via DocType definition
144	-- In doctype JSON: set "in_list_view" or "search_index" on fields
145	-- OR use hooks.py:
146	-- after_migrate = ["myapp.patches.add_custom_indexes"]
147	```
148
149	---
150
151	## Redis Configuration
152
153	### Memory Management
154
155	```conf
156	# /etc/redis/redis.conf (or bench config/redis_cache.conf)
157
158	# Set maximum memory — NEVER let Redis use all available RAM
159	maxmemory 512mb
160
161	# Eviction policy — allkeys-lru is best for cache use
162	maxmemory-policy allkeys-lru
163
164	# Disable persistence for cache Redis (performance boost)
165	save ""
166	appendonly no
167	```
168
169	### Fr

People who liked this also viewed

6 picks

Type