Index - Mirage

The index keeps directory listings and file metadata, so ls, find and stat answer from one listing instead of asking the backend each time.

`type`	Keeps it	Keys
`ram`	In the process’s memory	`ttl`, default `600` seconds
`redis`	On Redis, under the workspace’s own keys	`url`, `key_prefix`, `ttl`

The key is index, beside store. A mount’s own index replaces it for that mount.

For the whole workspace

YAML
Python
TypeScript

workspace_id: agent
index:
  type: redis
  url: redis://redis.internal:6379/0
  ttl: 300

from mirage import Workspace
from mirage.cache.index import RedisIndexConfig

ws = Workspace(
    mounts,
    workspace_id="agent",
    index=RedisIndexConfig(url="redis://redis.internal:6379/0", ttl=300),
)

import { Workspace } from '@struktoai/mirage-node'

const ws = new Workspace(mounts, {
  workspaceId: 'agent',
  index: { type: 'redis', url: 'redis://redis.internal:6379/0', ttl: 300 },
})

With the index on Redis, one listing answers every instance of the workspace. The keys belong to the workspace, and naming the same key_prefix in several workspaces shares one index, as for the file cache.

For one mount

A mount whose listings change rarely can keep them longer than the rest:

YAML
Python
TypeScript

mounts:
  /archive:
    vfs: s3
    config:
      bucket: agent-archive
    index:
      type: ram
      ttl: 3600

from mirage import Mount, Workspace
from mirage.cache.index import IndexConfig
from mirage.vfs.s3 import S3VFS, S3Config

ws = Workspace({
    "/archive": Mount(
        S3VFS(S3Config(bucket="agent-archive")),
        index=IndexConfig(ttl=3600),
    ),
})

import { Mount, S3VFS, Workspace } from '@struktoai/mirage-node'

const ws = new Workspace({
  '/archive': new Mount(new S3VFS({ bucket: 'agent-archive' }), {
    index: { type: 'ram', ttl: 3600 },
  }),
})