Help & technical support

Check delivery status, addresses and management access in the client area. Verify the plan, OS, console and remote access before configuring updates, access rules and backups. If a paid order is pending, provide its order ID rather than placing another order.

Distinguish a timeout, refusal and authentication failure. Check IP, port and account, then use the product’s available console to inspect boot, listeners and firewall rules. Windows also requires Remote Desktop to be enabled. Retain an existing session while testing access changes.

Check A/AAAA records, web listeners, certificates and application logs. Inspect proxy and upstream services for 502/504 and permissions for 403. With a CDN, separate edge and origin failures and provide the hostname, error and timestamp.

Include client location, ISP, destination, timezone-aware timestamps, both route directions and actual application results. Intermediate ICMP loss alone does not establish service loss. Use an address for the purchased product rather than an unrelated regional reference.

A reinstall may overwrite the system disk; consult the product’s stated retention behavior. Export databases, files and configuration to independent storage and verify recovery before confirming the target instance and disk. Reinstallation is not a routine update.

Local snapshots depend on the original storage and do not independently protect against its failure. Keep separate backups and test consistent restoration. Check your product’s backup availability and retention; do not assume automatic backups are included.

Provide the incident window, resource trends and recent changes. Review available memory, swapping and OOM events, and check disk bytes and inodes. Correlate queries, backups, logs and processes before resizing or interrupting critical services.

Check the host driver with nvidia-smi before device visibility in containers or Python. Provide device, driver, framework, model and errors, plus precision, context, batch and concurrency for OOM. Reproduce with a small workload; model-file size is not runtime memory usage.

Check invoice ID, payment time, amount and service state. Provide a verifiable transaction reference and redacted receipt for an uncredited payment instead of paying twice. Renewal and recovery follow product terms; service restoration does not guarantee expired data can be recovered.

Describe source and destination, data volume, dependencies and downtime tolerance, then confirm supported options. Validate backups and the destination, coordinate final sync and writes, and reconcile new data before rolling traffic back.

Separate delivery, instance access and infrastructure incidents from self-managed application issues. Diagnostics help establish ownership. Development, data repair, migration and managed operations depend on the service agreement; scope and completion time must be confirmed.

Confirm the task, scope and window through an official ticket and prefer a temporary least-privilege account. Do not expose secrets in public messages or screenshots. Back up before data changes and revoke temporary access and verify the service afterward.