CloudLinux

56% of hosting providers say customers leave over price. 41% say they leave for Wix or Shopify. Only one of those is about money. The 41% aren't shopping on price. They want a website, not a toolkit, and they'll trade flexibility for simplicity. A discount doesn't change what they're buying. The churn reasons you can fix directly: - Performance issues: 29% - Missing features: 25% - Security incidents: 19% - Poor support: 18% (All figures from the 2026 Web Hosting Trends Report) Each of these is an operating cost (tickets, escalations, cleanup hours) long before it becomes churn. Did the SaaS pull keep growing? What actually worked against it? The 2027 survey, run with WebPros , is open now. It takes a few minutes, and there's a $500 Amazon gift card prize draw. Take the survey 👉 https://hubs.ly/Q04yW5720 #WebHosting #HostingTrends #CloudLinux #WebPros

16 named Linux kernel exploits went public over 16 weeks this year, from late April to mid-August. Since then, five more CloudLinux security advisories landed, ranging from local root exploits to container escapes. Each puts hosting and ops teams in the same spot: patch now and risk a disruptive reboot, or wait for the maintenance window and stay exposed in the meantime. Neither holds up against an auditor or an attacker. Researchers now audit kernel code with AI at scale, so this pace isn't a blip. It's the new baseline. We track each one as it lands, CVE by CVE, with what it means for patch SLAs and compliance windows — in the CloudLinux newsletter. If you're the one explaining time-to-patch to an auditor, it's a worthy addition to your inbox. Subscribe: https://hubs.ly/Q04ypBfg0 What's your current time-to-patch on a critical kernel CVE? Curious how it compares.

Each isolated website on a CloudLinux server can now have its own CPU, memory, I/O and process limits. Per-site resource limits are Phase 3 of CloudLinux Isolates, now in beta. Give a site its own limits, and when it gets busy it's throttled on its own, while the other sites in the same hosting account keep the resources they depend on. What you get: - Sites can use spare capacity: site limits can add up to more than the account's, and the account's limit still caps the total - Per-site usage in lveinfo, lvechart and isolatectl stats, and fault emails that name the site that hit its limit - No more overhead than an account limit In beta for cPanel servers running CloudLinux 8 and 9, included in existing subscriptions at no extra cost. How it works and how to try it: https://hubs.ly/Q04yBwKS0 #CloudLinux #WebHosting

Node.js Selector can now keep one copy of each npm package version on a server and share it with every application that uses it. The new Shared node_modules Store is in beta. A package version one application has already installed isn't downloaded or stored again for the next. The server links the same files into each application's node_modules folder. What matters for hosting providers: - Less duplicate data on disk, fewer repeated downloads - Shared files don't count toward the account's disk space or inode quota - Account owners don't change anything: Run NPM Install and npm install work as before, on the Node.js version they selected Included in every CloudLinux license tier at no extra charge. How it works and how to get started: https://hubs.ly/Q04yyGkF0 #CloudLinux #NodeJS #WebHosting

Missed our session on the 2026 kernel exploit wave? The recording is up. In 45 minutes we covered: 16 named exploits on a timeline, and the future trend we expect What rebooting a fleet costs, from OVH's own post-mortem How KernelCare applies live patches while you keep serving traffic Concrete paths to a patched fleet, including the two-week free trial Watch on demand: https://hubs.ly/Q04yjbN50

MariaDB 10.6 hit end of life on July 6, 2026. Security patches stopped. Many hosting fleets are still running it. Here's the part that matters: 10.6 wasn't neglected. MariaDB shipped 79 fixes across 17 releases in five years. The last one, 40 days before EOL, closed a CVSS 10.0 vulnerability. Such a clean record comes from active maintenance, not sturdy software. Maintenance stopped at EOL. Exposure didn't. On a shared server, one unpatched database puts every site on it at risk, at once. For hosting providers running 10.6 across a fleet, migrating on the vendor's deadline means testing, downtime, and risk hitting every customer at the same time. With Endless Lifecycle Support (ELS) for MariaDB from CloudLinux and TuxCare, our sister company, there are two ways to handle it: - Keep 10.6 patched past EOL - Use ELS to properly prepare for migration Same version, standard yum/dnf updates, no forced downtime. Still running 10.6 somewhere? Learn more: https://hubs.ly/Q04xDLWj0 #EndOfLifeSupport #MariaDB #Security

In July, OVH rebooted tens of thousands of hosts running roughly 1M customer VMs to patch one kernel bug. Their own post-mortem says the procedure "will need to be repeated." The next KVM bug landed 18 days later. The fix those reboots delivered was on KernelCare's live patch feed by July 10. Running kernels got it without stopping. Tomorrow, live with Web Hosting Talk, we cover that gap: 45 minutes on the 2026 kernel exploit wave and how fleets stay patched without rebooting. Free, with live Q&A. Save your seat: https://hubs.ly/Q04y76-h0@

A kernel CVE drops on a Tuesday. On a mixed 40-server fleet, 24 AlmaLinux, 12 Ubuntu, 4 CentOS 7, that one CVE becomes 24 reboot notices and ~4,300 affected accounts. And a reboot calendar that takes two weeks to drain. The obvious fix doesn’t fix much: * kpatch → Red Hat * Canonical Livepatch → Ubuntu * Ksplice → Oracle Three distros. Three agents. Three consoles. And CentOS 7 still gets nothing because there’s no upstream fix left to backport. A cross-distro, rebootless approach changes the math: 36 of 40 servers patch in the next check cycle, no reboot. The four CentOS 7 machines get a backported fix through their existing package manager. No reboot. No maintenance notice. No reseller forwarding it. No midnight ticket. Because the real cost of a reboot isn’t the 90 seconds of downtime. It’s everything that follows. OCLC saw a 72% reduction in patching time and a 90% drop in vulnerability exposure window after switching to KernelCare and ELS. What does an unplanned reboot actually cost your support team in tickets? Full math + the 5-question checklist for evaluating patch tooling: https://hubs.ly/Q04y1w320

Email issues are one of the biggest support drains for hosting providers. In our 2026 Web Hosting Trends Report, 42% of hosters cited spam, blacklisting, and deliverability as a top resource drain, followed by CMS/application issues (39%) and performance and security incidents (35%). And the security risks behind the workload are clear: ✔️ 53% — outdated or vulnerable software ✔️ 51% — malware and ransomware ✔️ 40% — DDoS attacks ✔️ 36% — bot traffic ✔️ 34% — outbound spam and blacklisting Which of these is costing you the most time these days and what have you managed to automate? Our 2027 Industry Year in Review survey, by CloudLinux and WebPros , is now open. Take the survey 👉https://hubs.ly/Q04y1nGv0

We have 15 seats for Atlas Digital Summit 2026. Not 15 discounts — 15 seats, and that's the whole allocation. They're for the partners and customers we actually work with. The summit is private and invite-only: 28–29 September, Reykjavík. Two days of hosting executives on AI, strategy and the future hosting stack — in rooms small enough that you talk instead of listen. If Reykjavík is a maybe, claim a seat now and decide later. Registration link is in the first comment. Already going? Say so below — we'd rather book proper time with you than find you at the coffee station. #AISecurity #WebHosting #AtlasDigitalSummit