Filter by type or platform. New posts arrive every morning.
An ExternalTaskSensor pokes for hours while the upstream DAG finished long ago. Compare logical dates, fix execution_delta, set failed_states and timeouts, and know when to switch to assets.
Replicas stay at zero or one while the queue grows. Read the scale config, fix rule metadata and scaler auth, and tune min/max replicas so Container Apps scales with load.
Find why Azure Update Manager shows no assessment, failed installs or skipped maintenance runs, then fix patch mode, the agent, update sources and schedule assignments.
Retire an LLM app without leaving live keys, paid deployments, searchable indexes or prompt logs behind. Confirm there's no traffic, revoke access, delete in the right order and keep evidence.
A dbt test failed and the job stopped. Find the failing rows with store_failures, decide warn or error with severity, warn_if and error_if, filter with where, and rerun only what failed.
Retire a SQL Server database without breaking a hidden job or report. Check usage and dependencies, take it offline for a scream test, keep a verified final backup, then drop it and clean up logins, jobs and history.
Logins fail with clock skew errors, AWS calls return Signature expired or S3 says RequestTimeTooSkewed. Find the drifting EC2 instances and point chrony or w32time at 169.254.169.123.
A foreign catalog on Azure SQL times out, rejects the login, shows no tables or runs slowly. Check the network path from the compute, the connection credentials, Unity Catalog grants, metadata refresh and pushdown.
Requests to an LLM API sit for minutes before failing, and worker threads pile up. Set explicit connect and read timeouts, cap retries and size the connection pool in the OpenAI SDK and boto3.
Error 1205 means a statement waited longer than innodb_lock_wait_timeout for a row lock. Find the blocker with sys.innodb_lock_waits, see what it last ran, clear it safely and fix the access pattern.
An alarm or event fired but no email, Slack, PagerDuty or webhook message arrived. Check the subscription state, the publisher, the endpoint and policies in that order.
When an engineer leaves or changes team, remove their OCI access in the right order, revoke every credential type, and find the resources and automation that still depend on them.
AWS charges for every public IPv4 address, in use or idle. CLI commands to list Elastic IPs and auto-assigned public IPs, see the cost, release safely and stop new ones appearing.
Ready-to-run SQL against Unity Catalog information_schema to find tables and owners, search columns, list grants, tags, masks and row filters, and spot stale or external tables.
A DAG suddenly creates a run for every missed interval and hammers your warehouse. Pause it, find what triggered catchup, deal with queued and running runs without losing data, then repair any partial loads.
An AKS LoadBalancer Service never gets an IP. Read the Service events to find the cause, then fix cluster identity permissions, internal subnet annotations, public IP resource groups, quotas and policy denies.
A web app with regional VNet integration still times out or resolves a public IP for a private SQL, Storage or Key Vault endpoint. Test from inside the app's sandbox, then check DNS, routing, NSGs and the target firewall.
Learn Event Hubs by using it. Create a namespace and an event hub, grant access with Entra ID, send and receive with Python, save your position with checkpoints and read the metrics that show if it's healthy.
Learn how Azure RBAC decides who can do what. Understand principals, roles and scopes, grant access with the CLI, check what someone really has and remove access when it's no longer needed.
Create a VPC one step at a time so you understand each part. Add a public and a private subnet, an internet gateway and a route table, check what makes a subnet public, then clean it up in the right order.
Build a VCN one piece at a time so you understand it. Add an internet gateway, a NAT gateway, route tables, a security list and two subnets, then see what makes each subnet public or private and clean it up.
Read data from your own ADLS account with no secrets. Create an access connector, grant it a storage role, register a storage credential and external location, then grant access and read the files.
See what cosine, dot product and Euclidean distance actually measure with runnable Python, then choose the right metric for your embedding model, read scores correctly and avoid mistakes that hurt recall.
Create a small Linux VM with az vm create, connect over SSH, limit the SSH rule to your own IP, stop paying when it's idle and delete everything cleanly when you finish.
Stop building jobs by hand in the UI. Write a small databricks.yml with a notebook job and dev and prod targets, deploy it from the CLI, run it and see what development mode changes for you.
A selective filter reads almost every file in a Delta table. Measure pruning in the query profile, inspect file stats, then fix stats columns, filter expressions, data layout and how you test it.
A stream reading a Delta table stops after a MERGE, UPDATE or DELETE on the source. Find the commit that caused it, decide what the downstream must see, then choose skipChangeCommits, Change Data Feed or a rebuild.
A new ECS service revision never finishes while tasks start and stop in a loop. Read rolloutState, find why tasks stopped, separate container failures from target group failures, then add circuit breaker and alarm rollback.
An EKS Ingress never gets an ALB. Read events and controller logs, then fix the ingress class, subnet discovery tags, IAM permissions, webhook errors and target-type settings that block it.
Learn the handful of KQL operators you'll use every day. Find tables, filter by time, pick columns, count with summarize, chart a trend, join two tables and avoid the mistakes that make queries slow.
Launch a small Amazon Linux instance from the CLI. Create a key pair, allow SSH only from your own IP, find the latest AMI safely, connect, then stop or terminate it so you don't keep paying.
Acknowledged writes are missing after a primary failover. Confirm a rollback happened, find and review the rollback files, restore what's safe, then choose write and read concerns so that it can't happen silently again.
Take a consistent logical backup with mysqldump, compress it, restore it into a test database, check the result and fix the common access, GTID and definer errors you'll see along the way.
Learn how OCI access works by doing it. Create a compartment and a group, add a user, write readable policy statements with the right verbs and fix the NotAuthorizedOrNotFound error when access doesn't work.
Learn Object Storage by using it. Find your namespace, create a private bucket, upload and download files, share one file with a time-limited link, choose a storage tier and clean up safely.
Failed large uploads leave invisible parts behind. They add to storage costs and can stop a bucket being deleted. Measure them, review before aborting, resume what's still needed and add a lifecycle rule.
No posts match these filters yet.
Showing the latest 36 of 1464 posts. Open a platform above or search to see them all.