Dify backup and restore
What to back up for Compose and source deployments (database, storage, vector data, env files), consistent snapshots, offsite copies and a tested restore runbook.
Dify operations and troubleshooting
Day-two work on a self-hosted Dify: backups, cleanup jobs, queue monitoring, mail, scaling workers and fixing the failures operators hit most often.
Dify database cleanup and storage reclamation
Keeping PostgreSQL and storage from growing without limit: message and workflow run retention, embedding cache cleanup, orphaned file removal and VACUUM after large deletes.
Dify Celery workers and queue backlog monitoring
Scaling worker containers for indexing and workflow load, the dataset queue monitor with email alerts, and warm shutdown behavior during deploys.
Dify email delivery configuration
Configuring Resend, SMTP or SendGrid so invitations, password resets, ownership transfers, email codes and human input forms actually arrive.
Troubleshooting Dify outbound request failures
Diagnosing tools, HTTP nodes, model servers and external knowledge APIs that time out or return 403 from inside Dify: SSRF proxy blocks, private host allowlists, timeouts and proxy settings.
Troubleshooting Dify plugin installation failures
Fixing plugins that fail to install or run: signature errors, unreachable Marketplace or PyPI, package size limits, daemon timeouts and dependency mirrors.
Dify scheduled maintenance tasks
Which Celery Beat jobs to enable in production, from message and workflow run cleanup to unused dataset cleanup, plugin upgrade checks and the schedule poller.
Dify performance tuning for workflows and the API
Finding the bottleneck when apps slow down: database connection pools, Redis timeouts, worker counts, sandbox and SSRF pool limits, and HTTP node timeout ceilings.
Running several Dify instances on shared infrastructure
Hosting development, staging and production Dify instances on shared Redis, PostgreSQL and vector stores safely with key prefixes, index name prefixes and separate databases.
