MySQL 5.7 'Already Upgraded' and dpkg Interrupted Error
Fix 'This installation of MySQL is already upgraded to 5.7.29' and the interrupted dpkg error on Ubuntu by editing the postinst script.
TLS certificates now last 200 days and drop to 100 in March 2027. A practical guide to automating renewals with ACME and monitoring expiry before outages.

Earlier this autumn the first 200-day TLS certificates began to expire, so teams that used to renew once a year found they had to do it again only months later. Two more deadlines are close: Let's Encrypt's default certificates drop to 64 days on 10 February 2027, and the industry-wide maximum falls to 100 days on 15 March 2027. If renewal in your organisation still means a calendar reminder and someone copying files onto servers, TLS certificate automation is no longer optional. This is how I would approach it.
In April 2025 the CA/Browser Forum approved Ballot SC-081v3. Apple proposed it, and Google, Mozilla and Sectigo supported it. The ballot sets a stepped schedule for how long a publicly trusted certificate can be valid. It also limits how long a CA can reuse a domain control validation (DCV) before it must check again (Sectigo):
| From | Max certificate validity | DCV reuse |
|---|---|---|
| 15 March 2026 | 200 days | 200 days |
| 15 March 2027 | 100 days | 100 days |
| 15 March 2029 | 47 days | 10 days |
These rules cover every publicly trusted CA, not just one vendor, so changing supplier will not get you out of them. They do not cover private certificates issued by your own internal PKI, which matters later in this article.
Let's Encrypt is moving faster still. Its opt-in tlsserver profile has issued 45-day certificates since 13 May 2026. On 10 February 2027 the default classic profile changes to 64-day certificates with a 10-day authorisation reuse period. On 16 February 2028 it moves to 45 days with a 7-hour reuse period (Let's Encrypt). Let's Encrypt also stopped sending expiry reminder emails on 4 June 2025, so the old safety net has gone.
DigiCert expects the 100-day limit in 2027 alone to make manual procedures untenable (DigiCert). I agree with that.
The dates are easier to grasp with a real example. A certificate bought this month, in November 2026, is one of the last 200-day certificates you will get. It expires in early June 2027. Its replacement will be issued after 15 March 2027, so it can last no more than 100 days. From that point you are renewing every two to three months, and from 2029 roughly every month.
If you use Let's Encrypt and have not changed profile, your clock starts sooner. From 10 February 2027 every renewal gives you 64 days. Take a cron job that renews every 60 days, which is a common pattern. With a 64-day certificate that leaves a four-day margin, so one failed run causes an outage. With 45-day certificates the job would run after the certificate has already expired.
You cannot automate certificates you do not know about. Before changing any tooling, list every public hostname and where its certificate actually sits:
For each one, record the owner, the issuing CA, the expiry date and how it is renewed today. Certificate Transparency logs are useful here. Publicly trusted certificates are recorded in these public logs, so searching for your domains often turns up certificates nobody remembered ordering.
ACME is the standard protocol for requesting and renewing certificates automatically. Use a well-maintained client: Certbot or acme.sh on traditional servers, Caddy's built-in ACME support, or cert-manager on Kubernetes. A commercial certificate is no reason to stay manual. DigiCert, for example, offers ACME for DV, OV and EV certificates.
The most important design rule is to schedule a daily check and let the client decide when to renew. Do not hardcode "every N days". Clients should renew about two-thirds of the way through a certificate's lifetime (Cybersecurity News), and a good client works that out from the certificate itself. A typical daily job looks like this:
# Example: run daily; Certbot only renews certificates that are due
17 3 * * * certbot renew --quiet --deploy-hook "systemctl reload nginx"
Many distribution packages install a systemd timer that does the same job. Check whether one exists before you add a second.
Choose the validation method to suit the setup. HTTP-01 works well for a single web server. DNS-01 is a better choice for wildcards, hosts that are not reachable from the internet, and setups with several servers behind a load balancer. If you use DNS-01, give the client an API token limited to the records it needs, not your whole DNS account. DCV reuse also shrinks to 100 days in 2027 and 10 days in 2029, so validation must work every time without anyone stepping in.
Check for ARI support. ACME Renewal Information is now RFC 9773. It lets a CA tell your client to renew early, for example after a mass revocation. Let's Encrypt recommends it as a form of emergency preparedness. When you compare clients, prefer one that supports ARI.
A renewal only counts once the service is actually serving the new certificate. A common failure is a certificate that renews successfully on disk while the web server carries on serving the old one until it expires. To avoid that:
systemctl reload nginx, or the equivalent for Apache, HAProxy or your mail server)With reminder emails gone, your monitoring has to be your own and has to run separately from the renewal system. Check from outside, the way a visitor connects. Alert when roughly a third of a certificate's lifetime remains, and again at seven days. Here is a simple example check:
#!/usr/bin/env bash
# Example: days until a host's certificate expires (GNU date)
host="example.com"
end=$(echo | openssl s_client -servername "$host" -connect "$host:443" 2>/dev/null | openssl x509 -noout -enddate | cut -d= -f2)
days=$(( ( $(date -d "$end" +%s) - $(date +%s) ) / 86400 ))
echo "$host: $days days left"
Your existing monitoring platform can probably do this already. What matters is that someone receives the alert and acts on it.
Some systems cannot run an ACME client: older appliances, legacy servers and some hosted platforms. For each one, choose one of these routes:
From now on, add "supports ACME or a certificate API" to your procurement requirements for any device that serves HTTPS.
In my experience, many organisations here still buy annual certificates through resellers and install them by hand on web servers, firewalls and appliances. That model stops working in 2027, so it is worth fixing a few local issues now:
At my company, Standing, we help clients with cloud hosting, managed services and information security. Moving certificates from manual renewal to automated renewal with monitoring is part of that work, and we follow the same steps described here.
If you use Let's Encrypt classic, finish before 10 February 2027. Everyone else should finish before 15 March 2027.
| Where the certificate lives | Recommended approach | Key check |
|---|---|---|
| Standard web server | Certbot, acme.sh or Caddy with a daily check and a reload hook | New certificate is actually served |
| Kubernetes | cert-manager with DNS-01 or HTTP-01 | Ingress picks up renewed secrets |
| CDN or managed hosting | Use the provider's managed certificates or ACME integration | Who owns the DNS records |
| Network appliance or VPN | Proxy in front, vendor API, or lifecycle tool | Procurement requires ACME or an API |
| Internal-only service | Private CA with its own automation | Trust distributed to clients |
As a quick checklist: build the inventory, automate with ACME and ARI, add deploy hooks, set up independent expiry alerts, and assign a named owner for each item.
Shorter certificate lifetimes are not a problem in themselves. The problem is any manual step left in the process. If you remove those steps before March 2027, the next reductions will make no difference to you.
🤖 Researched and drafted with AI assistance, guided by my editorial standards. Spotted something wrong? Tell me.