Now Supporting .NET 10 LTS! | Specialist Blazor & ASP.NET Hosting on AWS
Each site runs in its own IIS application pool, and a watchdog we built checks every pool every 2 minutes. A stuck pool is rebuilt automatically, usually before you notice. Around it: guards for your SQL Server databases, server checks every 2 minutes, and a support agent that can make fixes at any hour.
Sometimes an IIS application pool gets stuck: requests queue up, but Windows never starts a worker process to serve them, and the site times out with no error anywhere. Our watchdog looks for exactly that every 2 minutes. When requests are still waiting with no worker in two checks 20 seconds apart, it rebuilds the pool under the same name with the same settings, and your site answers again. Each repair is recorded, so our team sees it and can tell you when it happened.
A pool stops starting worker processes. Visitors' requests queue inside Windows and time out, with no error in any log.
Every 2 minutes it reads the request queue of every pool. Requests waiting with no worker process, in two checks 20 seconds apart, means the pool is stuck.
It recreates the pool under the same name with the same settings and puts your site back on it. Your files, bindings and certificates are not touched.
The repair is logged with its time, so our team sees it and support can tell you exactly when it happened.
Checked every 2 minutes; a stuck pool is back within about 2 minutes, without a ticket.
The same system watches the rest of the stack. Each part fixes a failure we have seen on Windows hosting, and every plan has all of it.
IIS has a safety switch called rapid-fail protection: after several crashes in a short time it stops the application pool, and the site stays down until someone starts it by hand. Our watchdog starts those pools again within minutes. If an app keeps crashing even after that, the watchdog stops retrying, so a broken app is not restarted forever, and the recorded times help you find the cause.
By default IIS shuts down an idle application pool, and the next visitor waits while .NET starts from scratch. Our watchdog sets idle pools to suspend instead: the worker process is paged out but kept, so the first request after a quiet spell is answered much sooner. Pools that are serving traffic are never touched for this.
With AUTO_CLOSE on, SQL Server shuts a database down when its last connection closes and recovers it on the next login. On a busy server that shows up as slow first requests and intermittent "Unable to complete login process due to delay in prelogin response" errors. Plesk creates new databases with AUTO_CLOSE on, so our guard checks every hour and turns it off, a few databases at a time so nothing is disturbed.
When we first turned it off, login checks went from 9.5 seconds on average, with failures, to 20 ms with none.
Bots try to log in to every database server on the internet. Rather than closing remote access, which you need for SQL Server Management Studio and your own tools, our servers watch the login logs and firewall any address that keeps failing. Ordinary blocks lift after 24 hours. If your own address gets caught after a few mistyped passwords, our support agent can check and lift the block from your ticket.
Every 2 minutes our control panel requests each of our web server addresses and the Plesk login page. Two failed rounds in a row count as an outage: the team is emailed at once, and the recovery is recorded with how long it lasted. The same records answer your questions about a given time (below).
Our AI support agent answers tickets at any hour, from our knowledge base and your account's actual settings rather than guesses. Within your plan it can act, not just advise: give a database the size your plan includes, lift a database login block on your own address, turn on DKIM signing for your mail, raise a mailbox's sending limit a little, and show which filter or spam check handled a message. Anything outside those limits goes to a person, Monday to Friday, 9 AM to 6 PM Eastern.
When a site misbehaved at 2:05 PM, the useful question is whether anything happened on our side. The answer comes from the records above: the web server checks, your own application pools and the SQL Server service, stated as plain facts with times in your time zone. A recycled or rebuilt pool is reported as the automatic recovery it was, not as an outage.
Self-healing application pools, SQL Server 2025 & 2022, full Blazor support and FREE SSL, with a 30-day money-back guarantee for new accounts. Three plans from $9.49/mo.