Hardening, vulnerabilities, change and incidents
Practise these as flashcards. One or two lines each; the ones you miss come back first.
Start freeEvery person and system gets only the access it needs, only for as long as needed. Done with RBAC, groups instead of individual grants, regular access reviews, and PAM tools for admin access.
An incident is restoring service as fast as possible. A problem is finding and fixing the root cause so the incident doesn't repeat.
Standard changes are pre-approved for routine low-risk work. Normal changes go to CAB with steps, testing, rollback, risk and a window. Emergency changes handle active outages and get reviewed afterward. I present changes at CAB, answer blast-radius questions and close the ticket with the outcome.
Multi-factor authentication requires two of: something you know, have, or are. Passwords get phished, reused and leaked; a second factor stops most account takeovers.
A web application firewall inspects HTTP traffic and blocks attacks like SQL injection, XSS and bad bots, usually with OWASP rule sets. It sits on an L7 device: App Gateway, Front Door, F5 ASM, Cloudflare.
Find out what they actually need to do and grant only that: a sudoers rule for one command, a group membership, or a service account with specific rights. Document it.
One per app, no interactive login, long random passwords or managed accounts (gMSA, managed identity), an owner on record, rotation, and monitoring for use from unexpected hosts.
Run the rollback plan right away, confirm service is restored, communicate, then do a review: why did testing miss it and what changes in the process.
A write-up of timeline, impact, root cause, contributing factors and action items that focuses on systems and process, not on who made the mistake. People stay honest, so you actually learn.
Don't trust anything because it's on the internal network. Verify every request by identity, device and context, give least privilege, segment the network, and assume breach so you log and monitor everything.
Treat it as compromised: rotate it immediately, check logs for misuse, then remove it from history (and know that forks and clones may still have it). Move the secret to a vault and add secret scanning to the repo.
The JVM doesn't trust the certificate chain the server presented, usually because the internal or intermediate CA isn't in its truststore. Import the CA into the truststore the app actually uses (cacerts or a custom keystore) and restart. Don't disable validation.