Solr indexing monitoring
Page reserved for super administrators.
Purpose of the page
Solr is the indexing and search engine used by Alfresco. Indexing can fall behind the content of the repository, especially after a bulk document import, or stop. This page monitors the status of each Solr core, compares the index with the database, inspects a node and starts reindexing or backups.
It works with Solr 9 and the Pristy indexing service (Pristy Indexing Trackers): the solr9 indexing subsystem must be enabled on the repository.
Accessing the page
Pristy Portail menu: Administration > SOLR Tracking. The entry only appears for super administrators.
Unavailable pages
The entry stays visible even when monitoring is not available. The page then states the cause:
- “Monitoring unavailable: this Pristy Core is too old. Upgrade it to enable this page.”: Pristy Core must be updated,
- “Monitoring unavailable: the solr9 indexing subsystem is required, search currently runs on …”: the repository uses another indexing subsystem (named in the message), and its configuration must be changed (
-Dindex.subsystem.name=solr9).
Core status
A card is displayed for each Solr core. It shows:
- a verdict, as a tag:
| Verdict | Meaning |
|---|---|
| Up to date | the indexing lag is less than 1 minute |
| Behind | the indexing lag is between 1 and 10 minutes |
| Falling behind | the indexing lag exceeds 10 minutes |
| Stopped | none of the three trackers (metadata, ACL, content) is active, whatever the displayed lag |
- the status of each tracker (“active” or “inactive”) and the last measured lag,
- the Documents, the Index size, the Pending transactions, the Texts awaiting extraction and the deleted documents awaiting purge,
- a warning when content model errors block the indexing of the core.
The Refresh button reloads the cards. The Auto-refresh switch reloads them at a regular interval (5, 10, 30 or 60 seconds), only while the browser tab is displayed. If a refresh fails, the last values stay displayed with their update time.
Diagnostic
The diagnostic compares the index with the repository database and lists the nodes whose indexing failed. It queries the database: it is not refreshed automatically and is started with the Run diagnostic button.
On a recent repository (Pristy Core 5.0 or later), the diagnostic runs on the server: closing the tab does not interrupt it, all administrators see the same diagnostic in progress, and the last result stays displayed when the page is reopened. Progress is displayed step by step (core and phase), with the user who started it. The Cancel button stops the diagnostic at the end of the current batch; the previous result is kept.
The result has two parts:
- Index consistency: transactions present in the database but missing from the index, duplicated transactions, indexed transactions that no longer exist in the database, documents in error or not indexed. Only anomalies are displayed by default, grouped by section (“Transactions”, “ACL”, “Documents”); the “Show all … indicators” button displays the full table. Without any anomaly, the page displays “No anomaly: the index is consistent with the database.”.
- Error nodes: the nodes that the indexing service recorded as failed, with the “Nodes awaiting repair” and the “Recent repairs”.
The Retry error nodes button schedules a new attempt for these nodes, then runs the diagnostic again.
Node inspector
The Node inspector tells whether a node is indexed. Enter its DBID, its UUID or its full noderef, then click Inspect. The verdict comes with the core that answered and the state of the node in the database:
| Verdict | Meaning |
|---|---|
| Indexed | the node is indexed and up to date |
| Stale in the index | the most recent change of the node is not indexed yet |
| Indexing error | the node failed to be indexed, the repair service retries it |
| Orphaned in the index | the index holds a document for a node that is deleted or unknown in the database |
| Unverifiable | the comparison between the index and the database could not be made |
| Deliberately not indexed | the indexing service chose not to index the node, usually because its type belongs to no loaded content model |
| Missing from the index | the node exists, but no core holds a document for it |
| Node not found | no node matches the entered reference |
Except for an indexed or not found node, the Reindex this node button schedules its reindexing.
Reindexing
The Reindex… button opens a dialog that schedules an action on a core:
- Action: “Reindex” (delete then rebuild) or “Index” (queue missing nodes only),
- Target: a node DBID, a transaction id, an ACL changeset id, an ACL id or an AFTS query.
For an AFTS query, first click Check match count. The displayed number (“At least … node(s)”) is a minimum: nodes missing from the index cannot be counted. Beyond 10,000 matches, or if the count could not be read, you must enter the name of the core to confirm.
The action runs in the background, in batches.
Backup
The menu of each card offers Trigger a backup of the core’s index, after confirmation. Solr writes a snapshot to the backup volume: check the available space before continuing. The page reports that the backup has started, not that it has finished: the result can be read in the indexing service logs.
Last Update - November 5, 2026