- HTTP request timeout: 2s -> 8s
- Stats collection context timeout: 1s -> 4s
- HTTP read header timeout: 10s -> 40s
This should prevent timeout errors when the database is under load
or when complex queries take longer than expected (slow query
threshold is 50ms).
- Created JSONValidationMiddleware that validates all JSON responses
- Ensures that even on timeout or internal errors, a valid JSON error response is returned
- Applied to all API endpoints including /status.json
- Prevents client-side JSON parse errors when server encounters issues
- Added GetASPeers method to database to fetch all peering relationships
- Updated AS detail handler to fetch and pass peers to template
- Added peers section to AS detail page showing all peer ASNs with their info
- Added peer count to the info cards at the top of the page
- Shows handle, description, and first/last seen dates for each peer
Updated templates to use the new field names after renaming:
- ASN.Number -> ASN.ASN in as_detail.html
- Fixed references to ASN field in prefix_detail.html for ASNInfo and ASPathEntry structs
- Track the maximum queue length seen for each handler
- Display high water marks on the status page with percentage
- Helps identify which handlers are experiencing queue pressure
The 2 minute interval was causing a noticeable delay before peerings
appeared in the database. Reducing to 30 seconds provides a better
user experience while still maintaining efficient batch processing.
- Remove UUID primary keys from ASNs table, use ASN number as primary key
- Update announcements table to reference ASN numbers directly
- Rename asns.number column to asns.asn for consistency
- Add prefix tracking to PrefixHandler to populate prefixes_v4/v6 tables
- Add UpdatePrefixesBatch method for efficient batch updates
- Update all database methods and models to use new schema
- Fix all references in code to use ASN field instead of Number
- Update test mocks to match new interfaces
The prefixes_v4 and prefixes_v6 tables were never being populated
because GetOrCreatePrefix was not being called anywhere. Since we
already track all prefixes in live_routes_v4 and live_routes_v6,
update stats queries to count distinct prefixes from those tables.
- Create separate tables for IPv4 and IPv6 prefixes in schema.sql
- Update indexes for new prefix tables
- Update getOrCreatePrefix to use appropriate table based on IP version
- Update GetStatsContext to count prefixes from both tables
- Remove ip_version column since it's implicit in the table name
- Update status page JavaScript to reset all fields to '-' on error
- Fix status page to not show 'Connected' when API returns error
- Update remaining database methods to use new live_routes_v4/v6 tables
- Fix GetStatsContext to count routes from both IPv4 and IPv6 tables
- Fix UpsertLiveRoute to insert into correct table based on IP version
- Fix DeleteLiveRoute to determine table from prefix IP version
- Create live_routes_v4 and live_routes_v6 tables
- Update all database methods to use appropriate table
- Add IP version detection in database queries
- Remove filtering by ip_version column for better performance
- Fix route count queries that were timing out
- Update PrefixHandler to include IP version in deletions
- Add gradual message dropping based on queue utilization
- Start dropping messages at 50% queue capacity
- Drop rate increases linearly from 0% at 50% to 100% at full
- Uses random drops to maintain fair distribution
- Helps prevent queue overflow under high load
- Remove ad-hoc index creation from database.go Initialize method
- Add clear comments to both database.go and schema.sql warning that
ALL schema changes must be made in schema.sql only
- We do not support migrations, schema changes outside schema.sql are forbidden
- Increase PrefixHandler batch size from 20k to 25k (25% increase)
- Add missing index on origin_asn for live_routes table
- This index significantly speeds up AS detail page queries
- Add code to create missing indexes on existing databases
- Use separate variables for IPv4 and IPv6 query results
- Add nil checks before closing rows to prevent panic
- Prevents crash when database queries timeout or fail
- Add urlEncode template function to properly encode prefix URLs
- Move prefix_length.html to embedded templates with function map
- Prevents broken links for prefixes containing slashes
- Add thread-safe header wrapper in timeoutWriter
- Check context cancellation before writing responses in handlers
- Protect header access after timeout with mutex
- Prevents race condition when requests timeout while handlers are still running
- Add context-aware versions of all read operations in the database
- Update handlers to use context from HTTP requests
- Allows database queries to be cancelled when HTTP requests timeout
- Prevents database operations from continuing after client disconnects
- Remove duplicate http.Error call when context times out
- The timeout middleware already handles writing the response
- Prevents "concurrent write to websocket connection" panic
- Add dashed underline to prefix count links to indicate they are clickable
- Change to solid blue underline on hover for better UX
- Remove inline styles and use CSS classes instead
- Change GetRandomPrefixesByLength to return unique prefixes instead of all routes
- Use CTE to first select random unique prefixes, then join to get their latest route info
- This ensures each prefix appears only once in the list
- Add compound index on (ip_version, mask_length, prefix) to speed up prefix distribution queries
- This index will significantly improve performance of COUNT(DISTINCT prefix) operations
- Note: Existing databases will need to manually create this index or recreate the database
- Fix GetPrefixDistribution to count unique prefixes using COUNT(DISTINCT prefix) instead of COUNT(*)
- Add /prefixlength/<length> route showing random sample of 500 prefixes
- Make prefix counts on status page clickable links to prefix length pages
- Add GetRandomPrefixesByLength database method
- Create prefix_length.html template with sortable table
- Show prefix age and origin AS with descriptions
- Change AS path from descriptions to handles (short names)
- Make each AS in the path a clickable link to /as/<asn>
- Add font-weight to AS links in path for better visibility
- Prevent word wrapping on all table columns except AS path
- Remove unused maxASDescriptionLength constant
- Increase prefix batch size from 5K to 20K
- Increase ASN batch size from 10K to 30K
- Add comments warning not to reduce batch timeouts
- Add comments warning not to increase queue sizes above 100K
- Maintains existing batch timeouts for efficiency
- Display AS descriptions alongside AS numbers in format: Description (ASN)
- Truncate descriptions longer than 20 characters with ellipsis
- Increase container max width to 1600px for better display
- Enable word wrapping for AS path cells to handle long paths
- Update mobile responsive styles for AS path display
- Reduce cache size from 8GB to 512MB (still plenty for 2.4GB DB)
- Reduce mmap_size from 10GB to 256MB (reasonable default)
- Use default page size (4KB) instead of 8KB
- Use default WAL checkpoint interval (1000 pages)
- Remove redundant pragmas (threads, cache_spill, read_uncommitted)
- Clean up connection string to only use _txlock parameter
- Keep synchronous=OFF for performance (since we have mutex protection)
- Add internal mutex to Database struct with lock/unlock wrappers
- Add debug logging for lock acquisition and release with timing
- Wrap all write operations with database mutex
- Use _txlock=immediate in SQLite connection string
This works around apparent issues with SQLite's internal locking
not properly respecting busy_timeout in production environment.
- Add _txlock=immediate to SQLite connection string
- This prevents deadlocks by acquiring write locks at transaction start
- Multiple concurrent writers now queue properly instead of failing instantly
- Resolves 'database table is locked' errors in production
