1
0
Fork 1
mirror of https://github.com/thatmattlove/hyperglass.git synced 2026-04-17 21:38:27 +00:00
thatmattlove-hyperglass/docs/pages/configuration/config/structured-output.mdx
Wilhelm Schonfeldt 4a1057651f
feat: comprehensive IP enrichment and traceroute improvements
MAJOR ENHANCEMENTS:

IP Enrichment Service (hyperglass/external/ip_enrichment.py):
- Increase IXP data cache duration from 24 hours to 7 days (604800s) for better performance
- Fix critical cache refresh logic: ensure_data_loaded() now properly checks expiry before using existing pickle files
- Remove 'force' refresh parameters from public APIs and admin endpoints to prevent potential abuse/DDOS
- Implement automatic refresh based on file timestamps and cache duration
- Add comprehensive debug logging gated by Settings.debug throughout the module
- Clean up verbose comments and improve code readability
- Update configuration model to enforce 7-day minimum cache timeout

MikroTik Traceroute Processing:
- Refactor trace_route_mikrotik plugin to use garbage cleaner before structured parsing
- Only log raw router output when Settings.debug is enabled to reduce log verbosity
- Simplify MikrotikTracerouteTable parser to expect pre-cleaned input from garbage cleaner
- Remove complex multi-table detection, format detection, and deduplication logic (handled by cleaner)
- Add concise debug messages for processing decisions and configuration states

Traceroute IP Enrichment (traceroute_ip_enrichment.py):
- Implement concurrent reverse DNS lookups using asyncio.to_thread and asyncio.gather
- Add async wrapper for reverse DNS with proper error handling and fallbacks
- Significant performance improvement for multi-hop traceroutes (parallel vs sequential DNS)
- Proper debug logging gates: only detailed logs when Settings.debug=True
- Upgrade operational messages to log.info level (start/completion status)
- Maintain compatibility with different event loop contexts and runtime environments

Configuration Updates:
- Update structured.ip_enrichment.cache_timeout default to 604800 seconds
- Update documentation to reflect new cache defaults and behavior
- Remove force refresh options from admin API endpoints

MIGRATION NOTES:
- Operators should ensure /etc/hyperglass/ip_enrichment directory is writable
- Any code relying on force refresh parameters must be updated
- Monitor logs for automatic refresh behavior and performance improvements
- The 7-day cache significantly reduces PeeringDB API load

PERFORMANCE BENEFITS:
- Faster traceroute enrichment due to concurrent DNS lookups
- Reduced external API calls with longer IXP cache duration
- More reliable refresh logic prevents stale cache usage
- Cleaner, more focused debug output when debug mode is disabled

TECHNICAL DETAILS:
- Uses asyncio.to_thread for non-blocking DNS operations
- Implements process-wide file locking for safe concurrent cache updates
- Robust fallbacks for various asyncio execution contexts
- Maintains backward compatibility while improving performance

FILES MODIFIED:
- hyperglass/external/ip_enrichment.py
- hyperglass/models/config/structured.py
- hyperglass/api/routes.py
- hyperglass/plugins/_builtin/trace_route_mikrotik.py
- hyperglass/models/parsing/mikrotik.py
- hyperglass/plugins/_builtin/traceroute_ip_enrichment.py
- docs/pages/configuration/config/structured-output.mdx
2025-10-05 21:25:58 +02:00

224 lines
9.5 KiB
Text
Raw Blame History

This file contains invisible Unicode characters

This file contains invisible Unicode characters that are indistinguishable to humans but may be processed differently by a computer. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.

import { Callout } from "nextra/components";
## Structured
Devices that support responding to a query with structured or easily parsable data can have their response data placed into an easier to read table (or JSON, when using the REST API). Currently, the following platforms have structured data supported in hyperglass:
- Arista EOS
- FRRouting
- Huawei VRP
- Juniper Junos
- Mikrotik RouterOS/SwitchOS
When structured output is available, hyperglass checks the RPKI state of each BGP prefix returned using one of two methods:
1. **From the router's perspective** - Uses the RPKI validation state as determined by the router itself
2. **From an external RPKI validation service** - Queries an external service to determine RPKI state independently
For external validation, hyperglass supports two backends:
- **Cloudflare**: Uses [Cloudflare's public RPKI service](https://rpki.cloudflare.com/) via GraphQL API
- **Routinator**: Connects to your own [Routinator](https://github.com/NLnetLabs/routinator) RPKI validator instance
Additionally, hyperglass provides the ability to control which BGP communities are shown to the end user.
For devices with structured traceroute support (Arista EOS, FRRouting, Huawei VRP, Juniper Junos, and MikroTik RouterOS), hyperglass can enhance the output with IP enrichment data including ASN information, organization names, country codes, and IXP detection using offline data from BGP.tools and PeeringDB.
| Parameter | Type | Default Value | Description |
| :-------------------------------- | :-------------- | :------------ | :------------------------------------------------------------------------------------------------------------------------------------- |
| `structured.rpki.mode` | String | router | Use `router` to use the router's view of the RPKI state, or `external` to use an external validation service. |
| `structured.rpki.backend` | String | cloudflare | When using `external` mode, choose `cloudflare` or `routinator` as the validation backend. |
| `structured.rpki.rpki_server_url` | String | | When using `routinator` backend, specify the base URL of your Routinator server (e.g., `http://rpki.example.com:3323`). |
| `structured.communities.mode` | String | deny | Use `deny` to deny any communities listed, `permit` to _only_ permit communities listed, or `name` to append friendly names. |
| `structured.communities.items` | List of Strings | | List of communities to match (used by `deny` and `permit` modes). |
| `structured.communities.names` | Dict | | Dictionary mapping BGP community codes to friendly names (used by `name` mode). |
| `structured.ip_enrichment.cache_timeout` | Integer | 604800 | Cache timeout in seconds for IP enrichment data (minimum 7 days/604800 seconds). |
| `structured.ip_enrichment.enrich_traceroute`| Boolean | true | When `structured:` is present, enable IP enrichment of traceroute hops (ASN, org, IXP). This must be true for enrichment to run. |
| `structured.enable_for_traceroute`| Boolean | (when structured present) true | When `structured:` is present this controls whether the structured traceroute table output is shown. Set to false to force raw router output. |
| `structured.enable_for_bgp_route`| Boolean | (when structured present) true | When `structured:` is present this controls whether the structured BGP route table output is shown. Set to false to force raw router output. |
### RPKI Examples
#### Show RPKI State from the Device's Perspective
```yaml filename="config.yaml" copy {2}
structured:
rpki:
mode: router
```
#### Show RPKI State from Cloudflare's Public Service
```yaml filename="config.yaml" copy {2-4}
structured:
rpki:
mode: external
backend: cloudflare
```
#### Show RPKI State from a Custom Routinator Server
```yaml filename="config.yaml" copy {2-5}
structured:
rpki:
mode: external
backend: routinator
rpki_server_url: "http://rpki.example.com:8080"
```
<Callout type="info" emoji="">
**Routinator URL Format**
The `rpki_server_url` should be the base URL of your Routinator HTTP web API endpoint. This is typically different from the RTR port (3323). The URL should not include the `/validity` path as hyperglass will append this automatically.
</Callout>
### Community Filtering Examples
#### Deny Listed Communities by Regex pattern
```yaml filename="config.yaml" {5-6}
structured:
communities:
mode: deny
items:
- '^65000:1\d+$' # don't show any communities starting with 65000:1. 65000:1234 would be denied, but 65000:4321 would be permitted.
- "65000:2345" # don't show the 65000:2345 community.
```
#### Permit only Listed Communities
```yaml filename="config.yaml" {5-6}
structured:
communities:
mode: permit
items:
- "^65000:.*$" # permit any communities starting with 65000, but no others.
- "1234:1" # permit only the 1234:1 community.
```
#### Append Friendly Names to Communities
```yaml filename="config.yaml" {2-10}
structured:
communities:
mode: name
names:
"65000:1000": "Upstream Any"
"65000:1001": "Upstream A (all locations)"
"65000:1101": "Upstream A Location 1"
"65000:1201": "Upstream A Location 2"
"65000:1002": "Upstream B (all locations)"
"65000:1102": "Upstream B Location 1"
"65000:2000": "IXP Any"
```
### IP Enrichment Examples
<Callout type="info" emoji="">
**IP Enrichment Requirements**
IP enrichment is currently supported for traceroute outputs on supported platforms.
The system uses offline data from BGP.tools (1.3M+ CIDR entries) and PeeringDB for maximum performance and reliability.
</Callout>
#### Enable IP Enrichment for Traceroute
```yaml filename="config.yaml" copy {2-4}
structured:
# Ensure `structured:` exists to enable structured output. By default the
# structured table output is enabled when this block is present. To disable
# the structured traceroute table, set `structured.enable_for_traceroute: false`.
ip_enrichment:
enrich_traceroute: true
```
#### Enable IP Enrichment with Custom Cache Timeout
```yaml filename="config.yaml" copy {2-5}
structured:
ip_enrichment:
enrich_traceroute: true
cache_timeout: 172800 # 48 hours
```
#### Enable IP Enrichment for Traceroute
```yaml filename="config.yaml" copy {2-4}
structured:
ip_enrichment:
enrich_traceroute: true
cache_timeout: 86400 # 24 hours (minimum)
```
<Callout type="warning" emoji="⚠️">
**Performance Considerations**
- Initial cache loading may take 30-60 seconds on first startup
- Data is cached locally using pickle format for ultra-fast subsequent loads
- Cache files are stored in `/etc/hyperglass/ip_enrichment/`
- Minimum cache timeout is 24 hours (86400 seconds) to prevent excessive API usage
</Callout>
### Structured Traceroute Configuration
<Callout type="info" emoji="">
**Structured Traceroute Support**
Structured traceroute with rich metadata is available for:
- **Arista EOS**: Parses Unix-style traceroute output with hostname, multiple RTT support, and MPLS labels
- **FRRouting**: Parses Unix-style traceroute output with load balancing and multi-path support
- **Huawei VRP**: Parses Unix-style traceroute output
- **Juniper Junos**: Parses traceroute output with MPLS labels, multipath, and partial timeouts
- **MikroTik RouterOS/SwitchOS**: Parses multi-table format with statistics
When IP enrichment is enabled, traceroute hops are enhanced with ASN numbers, organization names, country codes, prefixes, and IXP detection.
</Callout>
#### Complete Structured Traceroute Setup
```yaml filename="config.yaml" copy {2-12}
structured:
rpki:
mode: external
backend: routinator
rpki_server_url: "https://rpki.example.com"
communities:
mode: name
names:
"65000:1000": "Transit Routes"
"65000:2000": "Peer Routes"
ip_enrichment:
enrich_traceroute: true
cache_timeout: 86400
```
#### Structured Traceroute with Cloudflare RPKI
```yaml filename="config.yaml" copy {2-9}
structured:
rpki:
mode: external
backend: cloudflare
ip_enrichment:
enrich_traceroute: true
```
#### Minimal Structured Traceroute (No IP Enrichment)
```yaml filename="config.yaml" copy {2-4}
structured:
ip_enrichment:
enrich_traceroute: false # Traceroute will show basic hop info without ASN/org data
```
<Callout type="warning" emoji="⚠️">
**IP Enrichment Dependency**
Without IP enrichment enabled:
- Traceroute hops will only show IP addresses and RTT values
- No ASN, organization names, or country information will be displayed
- AS path visualization will be limited or unavailable
- IXP detection will not function
For the full structured traceroute experience with rich metadata, `ip_enrichment.enrich_traceroute: true` is required.
</Callout>