til

Today I learnt

Architecture

A good architecture maximises the number of decisions NOT made.

Uncle Bod Martin - https://www.youtube.com/watch?v=NeXQEJNWO5w

All peripherals stuff (UI, Database, Framework) should be written as plugins to the Use Cases

Bash

Add up size of several files

ls -l | grep "2020-08-20" | awk '{s+=$5}END{print s}'

Mysql

Indexing

https://www.youtube.com/watch?v=HubezKbFL7E

If you want to learn more about database indexing, I really recommend https://use-the-index-luke.com/ The entire talk is basically the first few chapters from this site :)

Explain best:

Const / Eq_Ref (B-Tree / Binary search) Can only be used if uniqueness is guaranteed
Ref / Range B-Tree traversal to find starting point, then scans
Index Full Index Scan (no filter)
ALL (avoid at all cost) Full Table Scan

Even with an index on create_at Mysql will not use it because we are using a function: YEAR()

SELECT SUM(total)
FROM orders
WHERE YEAR(created_at) = 2013

Explain will use ALL index, no possible keys, no key.

Let's try giving it the precise range:

EXPLAIN SELECT SUM(total)
FROM orders
WHERE created_at BETWEEN '2013-01-01 00:00:00' AND '2013-12-31 23:59:59'

Now we have possible keys, but no key used.

EXPLAIN SELECT SUM(total)
FROM orders FORCE INDEX (orders_created_at_idx)
WHERE created_at BETWEEN '2013-01-01 00:00:00' AND '2013-12-31 23:59:59'

ALL => range

Add index on 'total' > no need to force the index

Explain 'Extra' > Using index

Using index ONLY is the best way

Inequality operators

Index stops at the first inequality operator

Ex: BETWEEN

Multi column indexes

Using from left to right => The column order in a index matters

Tables size

SELECT
    TABLE_NAME AS `Table`,
    ROUND((DATA_LENGTH + INDEX_LENGTH) / 1024 / 1024) AS `Size (MB)`
FROM
    information_schema.TABLES
WHERE
        TABLE_SCHEMA = "video_encoder"
ORDER BY
    (DATA_LENGTH + INDEX_LENGTH)
    DESC;

ngork

To serve a site from a vagrant using ngrok, from the host:

ngrok http -host-header=rewrite local.domain.net:80

PHP / FPM

Problem with too many requests / connections dropped

tail -f /var/log/php5-fpm.log

[18-Aug-2020 15:55:38] WARNING: [pool www] seems busy (you may need to increase pm.start_servers, or pm.min/max_spare_servers), spawning 8 children, there are 6 idle, and 23 total children

Edit /etc/php5/fpm/pool.d/www.conf

pm.max_children = 200

(was 100)

service php5-fpm restart

@Gaëtan si tu as des pbm pour les uploads, tu as deux paramètres à prendre en compte : 1 - est-ce que la sessions fpm n'est pas fermée avant la fin de l'upload (et ça tu pourras pas le changer : l'utilisateur à, de mémoire, 600 secondes pour faire l'upload... ou 300... enfin pas toute sa vie quoi) 2 - est-ce que la connexion tcp ne plante pas à cause de la ligne et de la taille de la data qui transite : et là aussi, tu peux rien y faire

salt

To list all managed minions:

salt-run manage.status

Or list accepted keys:

salt-key -L

pixmin/til

til