Project description
Currently we are essentially blind to what is happening in production. Logs are scattered across a dozen servers, each application writes to its own files in its own format, and when something breaks, we log in via SSH and manually grep through gigabytes of text. Usually we learn about a problem only after a customer reports it. We want to solve this properly: build a centralized logging and observability stack that collects logs and metrics from every server and service in one place with search, so the whole team sees the picture without logging into the machines themselves.
The target architecture is an ELK or OpenSearch cluster for logs plus Grafana for metrics and visualization. The contractor is expected to deploy log collectors on our hosts, normalize and parse incoming streams into structured fields, and design dashboards that are clear to both engineers and management. On top of this we need alerting: notify when error rate rises, when response delay crosses a threshold, when a service stops sending data, or when disk space or memory runs out. Alerts should come to the channels we already use, with appropriate thresholds so we don't drown in noise.
Storage and its cost matter to us. We ask you to configure index lifecycle and rotation so fresh data stays hot and searchable, and old logs move to archive or are deleted on schedule. Describe the architecture, parsing rules, and the procedure for connecting a new service to the pipeline so our team can continue to maintain and expand the system themselves. If you have already managed such a stack in production, tell us how you would scale it for our volumes.
— Collecting logs from Linux servers and containerized applications
— Structured parsing, indexing, and retention policy
— Grafana dashboards for errors, delays, and resources
— Alerts for error spikes, delays, and silent services
The target architecture is an ELK or OpenSearch cluster for logs plus Grafana for metrics and visualization. The contractor is expected to deploy log collectors on our hosts, normalize and parse incoming streams into structured fields, and design dashboards that are clear to both engineers and management. On top of this we need alerting: notify when error rate rises, when response delay crosses a threshold, when a service stops sending data, or when disk space or memory runs out. Alerts should come to the channels we already use, with appropriate thresholds so we don't drown in noise.
Storage and its cost matter to us. We ask you to configure index lifecycle and rotation so fresh data stays hot and searchable, and old logs move to archive or are deleted on schedule. Describe the architecture, parsing rules, and the procedure for connecting a new service to the pipeline so our team can continue to maintain and expand the system themselves. If you have already managed such a stack in production, tell us how you would scale it for our volumes.
— Collecting logs from Linux servers and containerized applications
— Structured parsing, indexing, and retention policy
— Grafana dashboards for errors, delays, and resources
— Alerts for error spikes, delays, and silent services
Administration Cash, Bank transfer
