DB Schema

Logs

Three models work together for log ingestion and archival.

LogEntry

Individual log entries stored in MongoDB. Archived to S3 in batches when threshold is reached.

Field	Type	Description
`id`	ObjectId	Primary key
`episodeId`	String	Parent episode
`timestamp`	DateTime	When log was created
`level`	String	`"info"` \| `"warn"` \| `"error"` \| `"debug"` \| `"fatal"`
`message`	String	Log message
`metadata`	Json?	Arbitrary metadata
`sequenceNumber`	BigInt	Auto-incrementing per episode (0, 1, 2, ...)
`createdAt`	DateTime	Created timestamp

Indexes: [episodeId, sequenceNumber], [episodeId, timestamp], [episodeId, level], [episodeId, createdAt]

Archived log chunks in S3. Each chunk contains up to 50k logs compressed as JSONL.gz.

Unique: [episodeId, chunkNumber]

Indexes: episodeId, [startTime, endTime]

Aggregated statistics and archival state per episode.

Field	Type	Description
`id`	ObjectId	Primary key
`episodeId`	String (unique)	Parent episode
`totalLogs`	BigInt	Total logs ever (default: 0)
`currentLogs`	BigInt	Logs in MongoDB, not yet archived (default: 0)
`archivedLogs`	BigInt	Logs archived to S3 (default: 0)
`totalChunks`	Int	Number of S3 chunks (default: 0)
`nextSequenceNumber`	BigInt	Next sequence to assign (default: 0)
`firstLogAt`	DateTime?	Earliest log timestamp
`lastLogAt`	DateTime?	Latest log timestamp
`currentMongoBytes`	BigInt	Size of current logs in MongoDB (default: 0)
`archivedS3Bytes`	BigInt	Size of archived chunks (default: 0)
`levelCounts`	Json?	e.g. `{ "info": 1000000, "warn": 50000 }`
`archiveThreshold`	Int	Archive when `currentLogs` reaches this (default: 50000)
`chunkSize`	Int	Logs per chunk (default: 50000)
`isArchiving`	Boolean	Lock flag for concurrent archival (default: false)
`lastArchivedAt`	DateTime?	Last archival timestamp
`updatedAt`	DateTime	Last updated

Indexes: currentLogs