Jump to content

Recommended Posts

Posted
28 minutes ago, Elfocrash said:

Let's see for how long

heh.

don't forget that i am stand alone, all those years  i had no help from you or from someone else who know how to programming,(i am not talking about my team) "the outer circle" ,i will be glad if someone else (including you) will help. :) as you know this is my hobby i am not doing it for money. 

Posted
13 hours ago, Tryskell said:

It's never bad to get some competition, so good luck. I hope those 3 "lost" years will translate to a boost of quality.

Thanks, alrthrough I would suggest you to update acis to that if you did not update it yet: https://app.assembla.com/spaces/l2hellas/subversion/commits/508

 

Posted (edited)
5 minutes ago, Nightw0lf said:

Thanks, alrthrough I would suggest you to update acis to that if you did not update it yet: https://app.assembla.com/spaces/l2hellas/subversion/commits/508

 

 

No reason to use 6 when there is 8, and if it's edited, it would be with the ConnectionPooler (= L2DatabaseFactory).

 

If you make a WORKING combo HikariCp + mysql connector, I commit it. It's not the connector which will boost performances.

Edited by Tryskell
Posted (edited)
2 hours ago, Tryskell said:

 

No reason to use 6 when there is 8, and if it's edited, it would be with the ConnectionPooler (= L2DatabaseFactory).

 

If you make a WORKING combo HikariCp + mysql connector, I commit it.

I have a hikari cp/mysqlconnector 5 patch from 2015 but internet says hikari is unstable i did not test it with 6 or later, here take a look

 

Edit to reply on your edit

2 hours ago, Tryskell said:

It's not the connector which will boost performances.

Actually theoretically is (this is the reason I tried to test hikari in this patch), asside the code and how much you use it if the base is "crap" everything works with the "crap", I am not sure how many classes use database connections in a project, but lets say 40% of total classes most of them are on server load so lets take this out of the performance so the communications with the database after server load will be 5% of total project classes this 5% is crucial for server stability and response to clients so it might look small but it will boost every server/client interaction of every client, what do you think about that?

From my tests back in the days late 2014 I noticed big performance difference with HDD ofc (now on SSD its just milisecs), talking about player actions that cause the use of database, with timings sometimes reach 1-3 seconds for lagging but 0.3 and 0.2 msecs less was what I had achieved with live version of hellas and the test version via local network (stupid but wire and wireless tests), to sum up I think the connector can boost performances.

Edited by Nightw0lf
Posted
2 hours ago, Tryskell said:

It's not the connector which will boost performances.

Well this is just wrong. Ofc the connector can boost performance in multiple ways.

Posted (edited)
2 hours ago, Tryskell said:

Make some graphs and come back later.

Don't need to. It's just facts. Libraries can have performance improvements by definition. I'm not saying this specific one has them, but it's undeniable they it can have them. It also means that performance can actually be worse after an update for various reasons, even though this is unusual.

Edited by Elfocrash
Posted
1 hour ago, Kara` said:

Hikari is problematic as fuck i replaced it in my older freya project after 3 hours of usage. There is no performance at all since SQL is already way too fast.

https://github.com/brettwooldridge/HikariCP/wiki/"My-benchmark-doesn't-show-a-difference."

Posted
6 hours ago, Elfocrash said:

So basically the inner-configuration is well structured in Hikari and it doesn't create a flood. In terms of speed there is no huge difference to switch right away as i can see.

Posted (edited)
5 minutes ago, Kara` said:

So basically the inner-configuration is well structured in Hikari and it doesn't create a flood. In terms of speed there is no huge difference to switch right away as i can see.

If that's what you understood, I can't help you. At this point it is known that Hikari is the way to go in terms of connection management and performance with Tomcat JDBC being a close second. c3p0 is considered legacy software, it's not really maintained actively anymore and everybody who's doing something serious in the industry has moved away from it.

 

On the other hand l2j is not considered something serious so c3p0 is good enough, because nobody really knows much about data in here.

Edited by Elfocrash
Posted
2 minutes ago, Elfocrash said:

If that's what you understood, I can't help you. At this point it is known that Hikari is the way to go in terms of connection management and performance with Tomcat JDBC being a close second. c3p0 is considered legacy software, it's not really maintained actively anymore and everybody who's doing something serious in the industry has moved away from it.

 

On the other hand l2j is not considered something serious so c3p0 is good enough, because nobody really knows much about data in here.

We said the same thing using different words. Seem like hikari is the way to go in terms of management of connection but speaking of speed the difference is small i don't see why you wrote "if thats what you understood". I didn't say it's trash i said back when i tried it, it had many problems. (in 2016 or so i can't remember exactly) and i didn't notice any difference. Maybe i didn't knew for the con handling performance and i just tried to compare speed.

Guest
This topic is now closed to further replies.



  • Posts

    • Some pictures and I uplade also some interface to be tested.  Interfaces to download (ofc compiled just to test): Protocol 166: https://drive.google.com/file/d/1_ZMfCJVplQg8uXnP3nn2w9OByIyKhqXV/view?usp=sharing Protocol 447 Classic: https://drive.google.com/file/d/1tG07r4GgdgCv_9IHdUROFfxe_yKqkno1/view?usp=sharing Protocol 542 Live: https://drive.google.com/file/d/1d9woGO5icCP6l9VT2tDpcqPaFtXYVGiG/view?usp=sharing Protocol 557 Live: https://drive.google.com/file/d/1s4HkuIUXtCWUO_qwZ9bX3JNjQ9VXlk6f/view?usp=sharing Protocol 557 Classic: https://drive.google.com/file/d/1xwaJh09wabvKlz80K_nbQq2JtNRnwhF-/view?usp=sharing Images:  
    • Come on kara you can do better
    • Tem esse npc no fórum, porém sem o efeito, alguém tem pra interlúdio com o efeito?
    • And whats the reason posting this here? GoldMembership in 2026 in MAXC.
    • Added the protection module to the demo.   DDoS Guard Pro v2.0 is a system protection module for PlayMMO CMS designed to reduce the load on the website during HTTP floods, bot activity, suspicious frequent requests, and attacks on individual pages or API methods. Unlike simple global limiters, DDoS Guard Pro v2.0 supports flexible rules based on routes and HTTP methods. This allows you to block the entire site in a targeted manner, rather than blocking the entire site equally. You can set up protection for specific areas of the site, such as login, registration, APIs, administration, forms, and other sensitive areas. What is the purpose of the module? DDoS Guard Pro v2.0 helps protect your site from basic L7 attacks at the HTTP request level. The module is useful when your site receives: frequent requests from a single IP address; HTTP page floods; login or registration form flooding; automatic requests from bots; URL scanning; frequent API requests; suspicious activity spikes; load on individual CMS methods or pages. The module helps to reduce the load on PHP and CMS by limiting suspicious activity before it starts to create a serious load on the site. Main features Per-route and per-method Rate Limit In the new version, protection is configured not only globally, but also according to specific rules. You can set limits separately for: GET; POST; PUT; PATCH; DELETE; ALL. This allows you to flexibly protect different parts of your website. For example: for the login page, you can set a strict limit; for registration, you can set a separate limit; for the API, you can set a limit for reading and a limit for changing data; for regular website pages, you can set a soft limit or not set a limit at all. This approach reduces the risk of accidentally blocking regular users and makes the protection more accurate. Flexible rule system The module supports setting rules in the following format: METHODS|PATTERN|LIMIT|WINDOW|BURST_LIMIT|BURST_WINDOW|BLOCK_SECONDS|IDENTITY|NAME Example of rules: POST|*login*|10|60|5|10|600|ip|login_post POST|*register*|8|60|4|10|600|ip|register_post GET|*api*|300|60|80|10|120|ip|api_get PUT,PATCH,DELETE|*api*|80|60|20|10|300|ip|api_write This allows you to specify exactly: which HTTP methods to protect; which URLs or URL patterns to consider; how many requests are allowed; over what time period; what burst limit to use;  how many seconds to block the offender;  by which ID to count the limit;  what the rule is called. Burst protection against sharp spikes  In addition to the regular request limit, the module monitors sharp spikes of activity.  This is useful when a bot makes many requests in a few seconds. In this case, the protection can be activated faster, without waiting for the overall limit per minute.  Burst protection is especially useful for: authorization pages; registration; API; search; data submission forms; administrative sections. Support for different types of requests DDoS Guard Pro v2.0 works not only with POST requests. The module can control: GET — regular pages, API requests, search; POST — forms, login, registration, data submission; PUT — updating data via API; PATCH — partial data update; DELETE — data deletion; ALL — all methods at once. This makes the module suitable not only for regular sites, but also for CMS with API, personal accounts, game panels and administrative actions. Limit storage: Redis, APCu and file fallback In the new version, the module supports several options for storing temporary data. Available modes: Redis; APCu; file fallback. The auto mode tries to use the most suitable option: Redis; APCu; file storage as a fallback. Redis or APCu are suitable for more efficient operation, while the file storage is left as a fallback option for simple hosting environments that do not have additional extensions. JSONL logging The module records protection events in JSON Lines format. Logs are saved in the following file: storage/logs/ddos_guard.jsonl This format is more convenient than a regular text log, because each event is stored as a separate JSON record. The logs can record the following information: event time; IP address; HTTP method; URL; name of the triggered rule; reason for blocking; number of requests; action status; user-agent; protection mode. The JSONL format is convenient for analysis by external tools, log agents, and monitoring systems. Prometheus metrics DDoS Guard Pro v2.0 adds an endpoint for receiving metrics in Prometheus format. Endpoint: /?ddos_guard_metrics=TOKEN The token is set in the module settings. Metrics allow you to track: the number of processed requests; the number of rule activations; the number of blocks; activity by limits; protection events; module status. This allows you to connect monitoring and configure alerts so that the administrator can see when suspicious activity starts on the site. LOG ONLY mode The module has a LOG ONLY mode. In this mode, DDoS Guard Pro does not block users, but only records events and potential triggers in the log. This mode is recommended to be used after installation, in order to first see which rules are triggered, and only then to enable the real blocking.  This helps to avoid too strict limits and random blocking of regular users.  Support for Cloudflare and proxy  The module supports working behind Cloudflare or another reverse proxy.  With proper configuration, it is possible to take into account the real IP of the user, and not the IP of the proxy server.  This is important for sites that use:  Cloudflare; nginx reverse proxy; load balancers; CDN; hosting proxy protection. Nginx-recommendations DDoS Guard Pro v2.0 contains an example nginx-config: modules/ddos_guard/nginx-ddos-guard-example.conf This allows you to use the module as an additional application layer of protection, and to move the main coarse limits to the nginx level. Recommended protection scheme: Cloudflare / nginx / firewall → DDoS Guard Pro → PlayMMO CMS This approach is more correct than trying to solve all problems only at the PHP level.
  • Topics

×
×
  • Create New...

Important Information

This community uses essential cookies to function properly. Non-essential cookies and third-party services are used only with your consent. Read our Privacy Policy and We have placed cookies on your device to help make this website better. You can adjust your cookie settings, otherwise we'll assume you're okay to continue..