- Separate search latency from missing or stale results.
- Capture comparable tests and correlate their timestamps with logs.
- Verify result accuracy and permissions after every approved change.
1. Separate slow searches from missing results
Ask the affected user to demonstrate one search that reliably shows the problem. Record whether the delay happens before results appear, while opening a result, or only when looking for recently added content. These are different paths through the service.
| Observed symptom | Investigate first | Useful evidence |
|---|---|---|
| Results are slow but correct | Query scope, result pagination, application requests and competing workloads | The same query, filters, result count and elapsed time |
| New or revised files are missing | Ingestion and indexing freshness, supported file processing and failures | Document creation/change time, processing state and searchable time |
| Content search is affected; property search behaves differently | The deployed full-text search path and its connectivity | Comparable content/property searches and relevant component logs |
| Results appear quickly; opening a file is slow | Content retrieval, rendering, storage and network path | Separate timings for search, file retrieval and viewing |
| Only certain users or screens are affected | Application customization, permissions, filters and user context | Affected and unaffected roles using equivalent test cases |
Use an account that is entitled to see the expected document. A result hidden by legitimate permissions is not evidence of a broken index. Keep test content and diagnostic evidence within approved support channels.
2. Capture a baseline the team can reproduce
Record the application, repository, user role, search terms, filters, result count and test time, including the time zone. Measure a small set of representative queries under comparable conditions; a single unusually fast or slow run is not enough to establish a pattern.
- Note whether repeated requests benefit from caching.
- Compare an affected time window with a normal operating period.
- List recent releases, infrastructure changes, imports, integrations and permission changes.
- Correlate application and platform logs using the same timestamps.
A timeline often gives the investigation a useful starting point. For example, a recurring morning slowdown may overlap with an import or scheduled job rather than indicate a permanent capacity shortage.
3. Review query scope and application behavior
Check whether the screen requests more results or properties than the user needs. Review filters, pagination, result limits and custom logic that makes additional requests for each result. A page displaying 25 items should not have to retrieve thousands of documents and their related data before it becomes usable.
Measure the application path as well as the query itself. Custom rendering, repeated lookups and client-to-server round trips can add time after the repository has responded. Treat each proposed change as a hypothesis, then test it with the same baseline.
4. Inspect the indexing path used by your release
Where document-content search depends on indexing, inspect the components deployed in your environment. Look for processing delays, failed items, repeated retries, unsupported content and connectivity errors. Compare when a document changed with when it became searchable.
Separate indexing freshness from query response time. An up-to-date index can still serve a slow query, and a quick query can return stale or incomplete results.
5. Check infrastructure and competing workloads together
Review resource pressure across the application, repository, database and search layers. Examine CPU and memory pressure, storage latency, free capacity, network delays and concurrent batch activity. Establish which layer is waiting and what changed during the affected interval.
A restart can remove evidence or provide only temporary relief. Database tuning, index rebuilds and service changes need an approved plan, appropriate testing and a recovery approach. Follow your operating procedures and the relevant product guidance rather than making several changes at once.
6. Verify both speed and correctness
After an approved change, repeat the original tests with comparable data, roles, workload and caching conditions. Confirm that the expected documents and versions are returned and that permissions still behave correctly. Check whether other workloads became slower.
Document the cause supported by the evidence, the change made, the before-and-after observations and the monitoring needed to identify recurrence. Agree acceptance criteria with the service owner; do not declare success from one faster query.
What should a Documentum diagnostic pack include?
- Component versions and a current architecture diagram.
- Reproduction steps, affected applications and user roles.
- Example queries, filters, result counts and observed timings.
- Relevant logs with timestamps and time zones.
- Recent changes, workload overlaps and indexing-freshness observations.
- Business impact, urgency, changes already attempted and their results.
This gives a support specialist enough context to investigate the incident without first reconstructing it. For broader planning, see our Documentum upgrade checklist.
Working checklist
Use this checklist to record your review. Selections stay in this browser view and are not submitted.
0 of 6 checked
Frequently asked questions
Is slow Documentum search always an indexing problem?
No. Broad queries, custom application logic, database workload, content retrieval and infrastructure pressure can produce similar symptoms. First identify which operation is slow and whether the returned results are correct.
Should we rebuild the full-text index first?
Not as a default response. Establish the deployed search architecture and gather evidence first. A rebuild can be disruptive and should follow release-specific guidance, a reviewed plan and appropriate recovery arrangements.
Why can users open documents but not find new ones?
Opening a known document and finding it through content search can use different processing paths. Check indexing freshness, failed processing, filters and the user’s entitlement to the document before choosing a fix.
Which response time should the team measure?
Measure user-visible elapsed time, then separate application, query, retrieval and viewing timings where possible. Use consistent queries, roles, result counts and workload conditions to compare results.
Can TechTiera help troubleshoot Documentum performance?
Yes. TechTiera can assess the relevant application and platform layers, review logs and workload patterns, and help define a tested remediation and support plan for the installed environment.



