Discord a trouvé une solution radicale pour gérer son app de ses morts qui bouffe 4 Go de RAM sous Windows 11 (consommation ressentie : 4 millions de Go) : La redémarrer automatiquement quand elle dépasse ce seuil.
Les champions quoi ! Plutôt que de corriger les fuites mémoires, ils ont tout simplement décidé d’intégrer un auto-relaunch dans l’app en douce pour qu’elle se relance toutes les quelques heures.
Donc, si votre Discord a tourné pendant au moins 1 heure, que vous êtes inactif depuis 30 minutes (pas de souris/clavier), et que vous n’êtes pas en vocal ou en visio, l’app se redémarrera automatiquement dès qu’elle atteindra les 4 Go de conso mémoire.
Bien sûr Discord présente ça comme une solution temporaire pendant qu’ils bossent sur le vrai problème, mais je trouve ça marrant de normaliser ce bug en ajoutant une fonctionnalité qui le “contourne” bruyamment.
Le pire dans tout ça, c’est que le problème vient d’une connerie technique assez basique. L’app Discord utilise une bibliothèque appelée “systeminformation” qui appelle PowerShell avec des commandes comme Get-WmiObject Win32_logicaldisk juste pour récupérer des infos système basiques. Mais comme ils passent par Powershell
plutôt que d’utiliser les API natives
de Windows, ça bouffe de la RAM comme un gros porc.
Comme vous, je me demande pourquoi Discord ne refait pas son app avec un framework plus léger ? Hé bien c’est simple : ils sont coincés ! Parce Discord est bâti sur Electron, et Electron c’est un framework qui embarque un Chromium complet salade tomates oignons dans chaque application. Ça permet aux devs web de créer des apps desktop avec du JavaScript, du HTML et du CSS , mais le prix à payer c’est une app qui pèse 85 Mo à l’installation et bouffe 200-400 Mo de RAM au démarrage.
En théorie Electron permet de créer une app cross-platform facilement. C’est-à-dire d’avoir un seul code pour Windows, Mac et Linux. Mais dans les faits, Discord maintient quand même du code spécifique par plateforme pour les notifications, l’overlay gaming, les raccourcis système, etc. Bref, ils ont tous les inconvénients d’Electron (RAM, taille) sans vraiment profiter de l’avantage du “write once, run everywhere”.
C’est vrai que réécrire Discord en natif coûterait des millions. Faudrait refaire toute l’interface, toutes les fonctionnalités, tous les systèmes de plugins et de thèmes. Surtout que pendant ce temps, l’équipe actuelle continue d’ajouter des fonctionnalités sur leur usine à gaz, ce qui creuse encore plus la dette technique. C’est le sunk cost fallacy version logicielle… en gros, ils ont tellement investi dans Electron qu’ils ne peuvent plus reculer, même si repartir de zéro serait probablement moins coûteux sur le long terme.
Pourtant, des alternatives à Electron existent. Tauri est devenu le framework préféré des devs qui veulent de la performance … On est à 2-3 Mo d’installeur, 30-40 Mo de RAM au repos et il utilise Rust et le webview natif du système plutôt que d’embarquer Chromium. Les apps sont donc légères, rapides, et consomment 10 fois moins de ressources.
Y’a aussi Flutter, React Native Desktop, Qt… des frameworks qui produisent des apps vraiment natives avec des performances dignes de ce nom. Visual Studio Code démontre qu’Electron peut être performant si on l’optimise correctement, mais ça demande un boulot monstre et malheureusement, Discord n’a clairement pas envie de mettre les moyens.
Le vrai problème n’est donc pas technique, c’est économique, car pour 1 dev natif, y’a 8 devs web . Electron permet d’embaucher des devs JavaScript pas cher plutôt que des devs C++/Rust/Swift qui coûtent une blinde… donc, sacrifier la RAM des utilisateurs coûte moins cher que payer des ingénieurs système. Et comme les PC ont maintenant 16-32 Go de RAM, ils se disent que 4 Go pour du chat en ligne, c’est acceptable. Lol.
Bref, tout ça pour dire que Discord normalise le “patch-as-a-feature”, et j’imagine que demain Slack, Teams et tous les autres vont faire pareil. En attendant, jetez un œil à Ripcord et Stoat pour ceux qui veulent un truc mieux.
WordPress, c’est bien. Mais WordPress qui injecte des scripts d’emojis, des styles Gutenberg, des shortlinks et 47 autres trucs dont vous n’avez pas besoin dans chaque page de votre site… c’est moins bien évidemment. Heureusement, Terence Eden, un dev qui en avait marre de voir son code source ressembler à un plat de spaghetti, a compilé une petite liste de tout ce qu’on peut virer .
Car WordPress a adopté une philosophie de type “Decisions, not options” (des décisions, pas des options) où en gros, au lieu de vous laisser choisir, ils décident pour vous de ce qui est bon pour vous. Un peu comme Macron ^^. Le problème c’est que leurs décisions incluent un tas de fonctionnalités dont la plupart des gens n’ont rien à faire 🥲.
Par exemple les emojis. J’sais pas si vous savez, mais WordPress charge un script de détection d’emojis et une feuille de style dédiée sur CHAQUE page de votre site. Pourquoi tant de haine ? Hé bien parce que si vous tapez :-) dans un article, WordPress veut le transformer en joli emoji. Sauf que si vous utilisez les vrais emojis Unicode (comme tout le monde en 2025), hé ce script ne sert à rien. Et il y a aussi le grand remplacement des emojis dans les flux RSS…. Bref, tout ça, ça dégage.
Ensuite y’a le formatage automatique avec wptexturize qui transforme vos guillemets droits en guillemets typographiques “comme ça”. Et mon préféré, capital_P_dangit qui remplace automatiquement “Wordpress” par “WordPress” avec le P majuscule. Oui, vous ne le saviez pas, mais WordPress corrige l’orthographe de son propre nom dans vos articles. Mais quelle bande de nazes ^^.
Gutenberg, l’éditeur de blocs que j’adore, injecte lui aussi ses styles globaux même si vous utilisez l’éditeur classique. Et c’est pareil pour les styles de la librairie de blocs et l’éditeur de widgets basé sur les blocs. Si vous êtes resté sur le Classic Editor comme beaucoup de gens, tout ça ne sert alors qu’à alourdir vos pages.
Côté métadonnées, WordPress ajoute aussi pleiiiiin de trucs dans le code de vos pages comme les shortlinks, le RSD (Real Simple Discovery, un truc d’il y a 20 ans), des liens vers les flux de commentaires, les liens JSON de l’API REST…
Aux chiottes toutes ces conneries !
Le script de Terence fait aussi sauter l’ajout automatique des tailles d’images (wp_img_tag_add_auto_sizes), les templates de pièces jointes, et les block hooks qui modifient votre contenu. L’idée c’est donc de reprendre le contrôle sur ce que WordPress génère, au lieu de le laisser décider tout seul.
Et grâce à son script, le site de Terence (sans Philippe) obtient d’excellents scores sur PageSpeed Insights , ce qui prouve que tout ce bloat n’est vraiment pas nécessaire. Son script PHP complet fait environ 190 lignes et il est dispo sur son GitLab , bien commenté pour que vous puissiez choisir ce que vous voulez garder ou virer.
Attention quand même, certaines de ces désactivations peuvent casser des fonctionnalités si vous les utilisez vraiment. Par exemple, si vous avez des plugins qui dépendent de l’API REST, la virer complètement serait une mauvaise idée. Même chose pour les blocks Gutenberg si vous utilisez cet éditeur. L’astuce c’est donc de tester chaque modification une par une et de voir ce qui se passe.
Amusez-vous bien et un grand merci à Terence !
When setting up a NAS, one of the most important and long-lasting decisions you’ll make is choosing the right RAID level. This choice directly impacts how much protection you have against drive failures, how much usable storage space you retain, and how long rebuilds will take when things go wrong. Among the most debated options are RAID 5 and RAID 6, both of which use parity for data protection but differ in how much risk they can tolerate. RAID 5 offers single-drive failure protection with better capacity efficiency, while RAID 6 provides dual-drive fault tolerance at the cost of more storage overhead and longer rebuild times. It’s worth noting that although you can graduate a RAID 5 into a RAID 6 later if your needs change, this is a slow and resource-heavy process. On the other hand, RAID 6 cannot be reversed back into RAID 5, so it’s a decision that requires careful planning from the outset. The balance of speed, safety, capacity, and risk tolerance will determine which configuration is truly best for your setup.
IMPORTANT – It is essential to understand that RAID, whether RAID 5 or RAID 6, should never be considered a true backup solution. RAID protects against drive failures, but it cannot safeguard you from accidental deletion, malware, hardware faults beyond the disks, or disasters like fire and theft.
If you are looking for simplicity, RAID 5 will usually give you the best balance of speed, storage efficiency, and cost, but it comes with higher risk. RAID 6 is slower to rebuild, consumes more usable capacity, and involves heavier parity calculations, but it provides a much stronger safety net against drive failures. For smaller arrays with modest drive sizes, RAID 5 can be entirely sufficient, especially when paired with reliable backups. However, as drive capacities continue to grow and rebuild times stretch into days, RAID 6 becomes more attractive because it can withstand the failure of two drives without losing the array. In essence, RAID 5 is about maximizing space and performance with a moderate level of safety, while RAID 6 is about maximizing resilience and peace of mind at the expense of capacity and speed. Choosing between them comes down to how valuable your data is, how large your drives are, and how much risk you are willing to tolerate during rebuild windows.
For systems with fewer than 8 bays, RAID 5 will usually be sufficient unless you are running especially large-capacity drives or operating at a business scale where data loss cannot be tolerated. Once you reach 8 bays or higher, RAID 6 should be seriously considered, as the chances of a second drive failing during a rebuild increase along with the overall storage pool size and the scale of potential loss. At 12 bays and beyond, RAID 6 is effectively mandatory, as relying on RAID 5 at that scale means gambling with too many points of failure and too much at stake if something goes wrong.
| RAID 5 | RAID 6 | |
|---|---|---|
| Pros | Higher usable capacity (only 1 drive lost to parity) | Dual-drive failure protection |
| Faster rebuild times | Much lower risk of catastrophic rebuild failure | |
| Lower cost per TB | Strong choice for very large drives (10TB+) | |
| Less parity overhead (better write speeds) | Safer for arrays with 6+ disks | |
| Widely supported and simple to manage | More reliable for mission-critical or archival data | |
| Cons | Vulnerable if a second drive fails during rebuild | Slower rebuild times |
| Higher risk of data loss with large drives | Higher cost per TB (2 drives lost to parity) | |
| Less safe for arrays over 6–8 disks | More computational overhead, slightly slower writes |
The initial creation of a RAID array, sometimes called synchronization or initialization, is one of the first differences you’ll notice between RAID 5 and RAID 6. A RAID 5 setup generally completes its initial build faster because it only has to calculate and assign a single parity block across the drives. RAID 6, by contrast, has to generate and distribute two independent parity values on every stripe, which increases the workload on the system. This means that on a fresh setup, RAID 6 will take longer to complete the synchronization process before the array is fully operational, though this is usually a one-time inconvenience at the beginning of deployment. For home and small office setups, this extra build time might not matter too much, but in larger systems with many terabytes of data, it can mean several hours or even days of extra initialization work compared with RAID 5.
The difference becomes more significant when a drive fails and a rebuild is needed. In RAID 5, the system only needs to reconstruct the missing data using the surviving disks and a single parity calculation, which usually makes recovery noticeably faster. RAID 6, however, must perform double parity calculations and restore both sets of parity information onto the replacement drive, extending the recovery window. On large modern HDDs where rebuilds can take dozens of hours, or sometimes multiple days, this extra time becomes a major factor. The trade-off is that RAID 6 offers much stronger resilience while this rebuild is in progress, because the system can continue to operate and survive even if another disk fails during the process. In other words, RAID 5 rebuilds faster but carries more risk, while RAID 6 rebuilds slower but provides a crucial safety margin during the vulnerable degraded state.
Here is a recent video (using the UniFi server platform) that talks about RAID 5/6 vs RAID 10 build times and parity from 777 or 404:
The most important factor when comparing RAID 5 and RAID 6 is how well they protect data when drives fail. RAID 5 uses single parity, meaning the system can survive one drive failure without losing data. However, if a second drive fails during the rebuild, the entire array is lost. RAID 6 adds dual parity, which allows the system to tolerate the loss of two drives simultaneously. This extra layer of protection is especially valuable during rebuild windows, which can take many hours or days on modern high-capacity HDDs. In practice, RAID 6 dramatically reduces the risk of catastrophic data loss, at the expense of slower rebuilds and less usable capacity. A subtle but often overlooked vulnerability is the issue of batch manufacturing. Many users buy multiple drives at once, often from the same supplier, meaning the disks may come from the same production batch. If there was a hidden flaw introduced during manufacturing, it is possible that more than one disk could develop problems around the same time. With RAID 5, this creates a dangerous scenario: a second disk failure during a rebuild results in complete data loss. RAID 6 provides a safety margin against these correlated failures by protecting the array even if two drives fail close together in time. Another major risk comes from unrecoverable read errors (UREs) that can occur during rebuilds. Because every sector of every remaining drive must be read to restore the lost disk, the chance of encountering a read error rises significantly with larger drives. In RAID 5, a single URE during rebuild can corrupt the recovery process, whereas RAID 6 has an additional layer of parity to compensate, making it much more reliable during rebuilds. This is especially important in arrays of 8 or more drives, where the probability of encountering at least one problematic sector grows. For users with large arrays or very high-capacity drives, RAID 6’s extra fault tolerance is the difference between a successful rebuild and complete data loss.
One of the clearest differences between RAID 5 and RAID 6 lies in how much usable capacity you end up with. RAID 5 only sacrifices the equivalent of a single drive’s worth of storage to parity, which makes it the more space-efficient option. In a six-bay system with 10TB drives, RAID 5 would deliver 50TB of usable storage, while RAID 6 would only provide 40TB. That 10TB difference can be substantial when you are working with large libraries of data such as media collections, surveillance archives, or backups. For users trying to maximize every terabyte of their investment, RAID 5 makes the most efficient use of available space. However, RAID 6’s higher storage overhead translates directly into a higher effective cost per terabyte. Since two drives are always reserved for parity, the total usable space is reduced, and the price you pay for storage per TB goes up. For small home users, this may feel like wasted potential, but the trade-off is the additional layer of fault tolerance. In environments where the cost of downtime or data loss far outweighs the cost of an extra disk, RAID 6 provides stronger long-term value despite the higher price per terabyte. Ultimately, the decision comes down to whether you are more concerned with minimizing cost and maximizing space, or ensuring redundancy and peace of mind.
Some users prefer to run RAID 5 with a dedicated hot spare drive rather than choosing RAID 6 outright. In this setup, a single extra disk sits idle until one of the active drives fails, at which point the spare is automatically used for the rebuild. This reduces the amount of time the array spends in a degraded and vulnerable state, since the rebuild begins immediately without waiting for a replacement disk to be manually installed. While this approach still leaves you with only single-drive fault tolerance, it can feel like a middle ground between RAID 5 and RAID 6. In terms of capacity, RAID 5 with a hot spare sacrifices the same amount of usable space as RAID 6, but it does not provide the same dual-drive protection. For arrays of six to eight drives, this compromise can make sense if you prioritize capacity efficiency and faster automated recovery, but once you move into larger-scale storage systems, RAID 6 remains the safer and more resilient option.
When choosing between RAID 5 and RAID 6, the decision comes down to weighing efficiency against resilience. RAID 5 is faster to rebuild, provides more usable storage, and costs less per terabyte, which makes it well suited to smaller NAS setups or users who prioritize capacity and speed. RAID 6, on the other hand, offers stronger protection against drive failures, making it far more reliable for larger arrays and higher-capacity drives where rebuild times are long and risks multiply. The general consensus is that RAID 5 can still be a smart choice for arrays under eight bays, but RAID 6 becomes the clear recommendation for systems of eight drives or more, and an essential requirement at twelve drives and beyond. Above all else, it is critical to remember that RAID is not a backup. Neither RAID 5 nor RAID 6 will protect you against accidental deletion, ransomware, hardware faults beyond the disks, or disasters such as fire or theft. RAID is a safety net that improves availability, but it must always be paired with a proper backup strategy if your data truly matters.
SUBSCRIBE TO OUR NEWSLETTER Join Inner Circle Subscribe
This description contains links to Amazon. These links will take you to some of the products mentioned in today's content. As an Amazon Associate, I earn from qualifying purchases. Visit the NASCompares Deal Finder to find the best place to buy this device in your region, based on Service, Support and Reputation - Just Search for your NAS Drive in the Box Below
Need Help?
Where possible (and where appropriate) please provide as much information about your requirements, as then I can arrange the best answer and solution to your needs. Do not worry about your e-mail address being required, it will NOT be used in a mailing list and will NOT be used in any way other than to respond to your enquiry.
[contact-form-7]
TRY CHAT
Terms and Conditions
Ko-fi or old school Paypal. Thanks!To find out more about how to support this advice service check HEREIf you need to fix or configure a NAS, check Fiver
Have you thought about helping others with your knowledge? Find Instructions Here
|
|
|
While using a computer, having an eye on your performance is essential. Being conscious of your performance potentially helps you avoid problems connected with overload. So, in this guide, we will explore how to show the Windows 11 Performance Overlay.
Windows 11 Performance Overlay is a tool built into the computer for real-time system performance monitoring. This utility shows different metrics essential in understanding how your computer handles a range of tasks.
Some common metrics to find in Performance Overlay include the following:
There are many advantages to viewing or using the Performance Overlay. However, at the top of the list is that you get enough information to help you troubleshoot issues that result in lags or crashes on your computer.
Because overlay is often transparent, it does not obstruct your workflow or gameplay.
Xbox Game Bar is built into Windows 10 and 11. It is essentially a customizable gaming overlay. This utility gives access to essential functionalities without you leaving the game. You may show this performance overlay by following the steps below.
1. Press Windows + G to open the Game bar.
2. Click the Performance tab to open performance options.
3. Select your Performance options menu.
4. Tick the checkboxes for any information you want to display.
5. Return to the Performance window and click the Pin icon.
6. Latsly, press the Windows + G keys to hide your Game Bar.
On Windows 11, the Settings app allows you to tweak app and operating system functions. You may access the Windows Overlay option for your game bar from Settings.
1. Press Windows + I to open the Settings app.
2. On the left pane, click Gaming; on the right, click Game Bar.
3. Toggle the switch to Allow your controller to open Game Bar.
4. Now, you can use a gaming controller to open the game bar and configure Performance Overlay, as shown in the first solution.
All your performance data is shown in the Task Manager. This utility also has an Always on top feature that permanently displays your performance on the screen.
1. Press the Ctrl + Shift + Esc keys to open the Task Manager.
2. Click Settings at the bottom left, then under the Windows Management category, tick Always on top.
3. Click the Performance tab, then double-click CPU for a summary view of your performance.
Your performance is displayed at all times in the Task Manager. Simply launch it by pressing Ctrl + Shift + Esc, then click on the Performance tab on the left pane.
If you have read through this guide, you should now be able to view your performance on the operating system easily. Showing performance is one of the better settings for gaming on Windows 11.
Do you have further questions on Performance Overlay? Let us know in the comment section below.
No, they are not, even though they are related. The FPS counter only shows the number of frames rendered per second, while the Performance Overlay offers a wide range of performance metrics.
Enabling this functionality has a negligible impact on performance. In most computers, you will barely notice a difference.
The post How to Show Windows 11 Performance Overlay appeared first on Next of Windows.
Connexion