Running the histogram before the search is the cheaper order — bucket by hour, find the ten minutes where the error count jumps, then re-run the same filter with a narrow start and end time instead of paging through a day. Know the boundary before you rely on it: queries are filter strings, and the author notes the server does not support structured-data search.
A reader for SolarWinds Observability logs. One tool returns matching log entries, the other counts the same query into time buckets so a spike is visible before you read a single line.
- Log search over a filter string, narrowed by group, entity and an ISO 8601 start and end time, defaulting to the last 24 hours
- Results formatted with timestamps, hostnames and messages, paginated by page size and direction so a long window can be walked in either direction
- A histogram of the same query bucketed by minute, hour or day, in UTC or local time
- Histogram output as an ASCII chart to read in place, or as JSON with time ranges, counts, total and the query parameters, for a client that draws it
A SolarWinds Observability API token in SOLARWINDS_API_TOKEN, set in the client config, the environment, or a local .env file. It starts over stdio as npx -y mcp-solarwinds. Searches default to the last 24 hours and a page size of 50. On Windows, a spawn ENOENT on npx is worked around by pointing the command at node.exe and npx-cli.js by full path.
One command plus a key — npm install -g mcp-solarwinds, then supply credentials