Troubleshooting Common Issues in mnoGoSearch Pro StandardmnoGoSearch Pro Standard is a powerful, enterprise-ready search engine solution designed to index and search large volumes of content efficiently. While stable and feature-rich, administrators occasionally run into issues that can affect indexing, search relevance, performance, or availability. This article walks through common problems, how to diagnose them, and practical steps to resolve each one.
1. Indexing Problems
Symptoms:
- New documents aren’t appearing in search results.
- Partial or incomplete indexing.
- Indexing jobs fail or get stuck.
Common causes and fixes:
-
Permissions and file access
- Ensure the mnoGoSearch daemons (mnogod, mnogosearch) run under a user with read access to the data sources and write access to the index directory.
- Check filesystem ownership and permissions; fix with chown and chmod as needed.
-
Configuration errors
- Verify your document source configurations (e.g., file lists, web crawl settings, database connectors) are correct. Typos or wrong paths prevent documents from being discovered.
- Confirm the indexing configuration files (like mnoGoSearch.conf and site-specific cfg files) have the right settings: charset, parser paths, and fetch limits.
-
Resource limits
- Large indexing jobs can exhaust memory or CPU. Monitor system resources; consider increasing available RAM, CPU, or using smaller batching.
- Adjust mnoGoSearch’s process limits (threads, fetchers) to match server capacity.
-
Corrupted index
- If the index is corrupted, rebuild it. Stop the search daemons, remove or move the index directory, and run the indexer from scratch:
mnogod --stop mv /var/lib/mnogosearch/index /var/lib/mnogosearch/index.bak mnogod --start muchindex --reindex - Always back up indexes before deletion.
- If the index is corrupted, rebuild it. Stop the search daemons, remove or move the index directory, and run the indexer from scratch:
-
Connector-specific issues
- For web crawls, ensure the crawler respects robots.txt and isn’t blocked by rate limits or CAPTCHAs.
- For database connectors, verify DB credentials, network access, and that queries return expected rows.
2. Search Relevance Problems
Symptoms:
- Irrelevant or low-quality search results.
- Expected documents rank too low.
Diagnosis and solutions:
-
Analyze scoring parameters
- mnoGoSearch uses configurable ranking factors. Review weightings for term frequency, document popularity, and field boosts in the ranking configuration.
- Increase boosts for important fields (title, URL, metadata) to prioritize them.
-
Tokenization and charset issues
- Ensure correct charset handling (UTF-8 vs. legacy encodings). Misconfigured charsets can split or misinterpret words, harming relevance.
- Check tokenization rules and stemming settings—different languages require appropriate stemmers.
-
Stopwords and synonyms
- Review stopword lists—overly aggressive lists can remove meaningful query terms.
- Configure synonym maps where appropriate so related terms return consistent results.
-
Freshness and recency
- If recent documents are more important, add freshness decay or promote recently indexed items.
-
Field indexing
- Ensure key fields are indexed and stored appropriately. Fields marked as non-indexed won’t contribute to relevance scoring.
3. Performance and Latency Issues
Symptoms:
- Slow query responses.
- High CPU or memory usage during queries.
- Timeouts under load.
Troubleshooting steps:
-
Monitor system metrics
- Use top, htop, vmstat, iostat to identify bottlenecks (CPU, IO wait, memory).
- Check disk throughput; indexes are IO-intensive—consider faster disks (NVMe) or RAID tuning.
-
Cache tuning
- mnoGoSearch supports caching of query results and index structures. Increase cache sizes if memory allows.
- Ensure OS-level disk caches are not being constrained by swap or low memory.
-
Query optimization
- Complex queries, large wildcard searches, or inefficient filters can slow responses. Simplify queries or use precomputed facets.
- Limit result window sizes and use pagination.
-
Concurrency settings
- Tune thread counts, worker processes, and connection limits to match hardware. Too many threads cause contention; too few underutilize CPU.
-
Network considerations
- If search clients access the server over the network, check latency and bandwidth. Use compression (gzip) for responses if supported.
4. Crashes and Stability
Symptoms:
- mnoGoSearch daemons crash or exit unexpectedly.
- Frequent restarts or memory leaks.
Investigation and fixes:
-
Check logs
- Review system logs (/var/log/syslog) and mnoGoSearch logs for backtraces or error messages. Enable verbose logging temporarily to capture more details.
-
Run under a debugger or use core dumps
- Enable core dumps (ulimit -c unlimited) and analyze with gdb if crashes persist. Capture stack traces to pinpoint faulty modules.
-
Memory leaks
- Monitor process memory over time. If memory usage grows unbounded, identify modules or connectors causing leaks. Update to latest patches where known leaks are fixed.
-
Update software
- Ensure you’re running a supported mnoGoSearch Pro Standard version. Apply vendor patches; many stability issues get resolved in updates.
-
System limits
- Ensure ulimits and kernel parameters (file descriptors, mmap limits) are sufficient for the workload. Increase /etc/security/limits.conf and sysctl settings as needed.
5. Authentication and Access Issues
Symptoms:
- Users can’t access search UI or APIs.
- Authentication failures with SSO/LDAP.
Common causes and resolutions:
-
Configuration mismatches
- Verify authentication backend settings (LDAP server, SAML endpoints, client secrets). Test connectivity with ldapsearch or SAML test tools.
-
Time sync
- SAML and JWT rely on accurate time. Ensure NTP is configured and servers’ clocks are synchronized.
-
SSL/TLS problems
- Expired or misconfigured certificates lead to failed connections. Check certificate validity and chain; include intermediate certs.
-
Permission mappings
- Confirm user group mappings and ACLs in mnoGoSearch match expected roles.
6. API and Integration Problems
Symptoms:
- API calls return errors or unexpected responses.
- Connectors fail to deliver data.
Steps to resolve:
-
Inspect request/response
- Use curl or Postman to replicate API calls; check HTTP status codes and response bodies.
-
Validate schemas
- Ensure data sent to APIs conforms to expected JSON/XML schemas and field types.
-
Rate limits and throttling
- Some integrations may be rate-limited. Implement exponential backoff and retry logic.
-
Version mismatch
- Ensure client libraries and server APIs are compatible; check version notes for breaking changes.
7. Logging and Diagnostics
Recommendations:
- Enable appropriate log levels for troubleshooting, then revert to normal to avoid log bloat.
- Centralize logs (rsyslog, journald, ELK) for easier correlation across components.
- Use monitoring tools (Prometheus, Grafana) to track metrics: query latency, indexing throughput, CPU/memory, error rates.
8. When to Contact Support
Contact vendor support when:
- Crashes produce stack traces you can’t interpret.
- Critical data loss or widespread index corruption occurs.
- You hit undocumented errors or regressions after updates.
Provide support with:
- mnoGoSearch version, OS, and hardware specs.
- Relevant log extracts, core dumps, and reproduction steps.
- Index configuration snippets and sample documents causing the issue.
Quick checklist (summary)
- Check permissions and file access.
- Verify configuration files and connectors.
- Monitor and tune resources and caches.
- Rebuild corrupted indexes.
- Inspect logs and enable verbose logging for diagnostics.
- Keep software up to date and apply vendor patches.
Leave a Reply