SQLSTATE[HY000] [2002] Connection refused on any query (e.g. select * from admin_users where id = 2 limit 1), every day around 12:00 server time, for about one minute, then recovers by itself. Seen on 11, 12, 14, 15 and 16/10/2026 (e.g. 16/10 12:00:21). DB host: sql102.infinityfree.com.
Other Information
Credentials are unchanged and correct (works immediately before and after). This is not a max_user_connections (1226) error and we use no persistent connections — the TCP connection itself is refused, which points to the DB server being briefly unreachable/restarting rather than our code. Is there a daily maintenance/restart window on this DB server around 12:00? Other accounts on the same server seeing this?
Welcome to the InfinityFree community forum! Good to have you here.
Most people answering here are volunteers. They’re often the fastest source of help. Admin (the only staff member) checks in about once a day for issues on our end.
Volunteers can’t see your account or website. Share what you actually see, like error messages and what happens when you try something. That’s more useful than your theory of what’s wrong.
AI tools often get InfinityFree wrong. If a volunteer’s advice differs from what an AI told you, please try the volunteer’s suggestion first.
Solving an issue is a joint effort. Try the suggestions you get and report back what happened.
Beep boop, I’m a bot! I dug through the forum to get you started. I sometimes get things wrong though, so do with this what you will until a human has a look.
I can’t find anything on the forum or in the knowledge base about a daily restart or maintenance window on sql102 around 12:00. So I can’t confirm or rule out that this is the DB server, and I won’t guess.
You’ve ruled out the usual suspects well: correct credentials, not a 1226, and no persistent connections. Whatever is behind a one-minute refusal at the same time every day is on the server side. That means a person on our staff needs to look at sql102, and nothing you can change in your code will fix it.
One detail would help whoever looks: the exact time of day, with timezone, for a few of those failures. A single error that shows a clear pattern helps more than a general description. Please also say whether phpMyAdmin or other sites on the same account fail at the same moment. If phpMyAdmin fails too, that would point even more strongly at the server.
I will test phpMyAdmin + a second site on the same account during the next window (tomorrow ~11:59–12:01 local) and report whether they fail simultaneously.
This is the first time I’ve heard about this, and I’m surprised that you were able to measure this.
Based on what I see checking the MySQL status information, it seems that the MySQL server is restarted automatically every day at the same time. Given the timing, this must be something intentional, like an update or a cleanup operation. Unfortunately, I don’t know the exact details.
Please be aware that keeping server software up to date is critical for security, and that restarting a process regularly is necessary to make that happen, so obtaining 100% uptime is impossible if you also want to keep a system secure.
I wouldn’t worry too much about this. Even a minute of downtime per day puts you well above 99.9% uptime, which is very respectable for any hosting service.