Troubleshooting
Running redundant InTouch historians with parallel tag logging
Short answer
InTouch has no dual-Historian target, so a partner Historian configured for redundancy stays empty. Log process data in parallel: the same tag list imported into both Historians, an IDAS on each node reading the OI Server. Handle historical alarms separately with A2ALMDB, because IDAS failover does not carry alarm data.
On this page
- What the empty partner Historian looks like
- InTouch has no dual-Historian target
- Feed each Historian through IDAS, not InTouch
- Historical alarms need their own path
- What the alarm client does on failover
- Building the parallel-logging configuration
- Checking the bundled 5,000-tag limit for a 4,200-tag application
- Licensing the redundant pair
- Questions asked
A second Historian configured as a partner stays empty when InTouch is the only data source, and nothing in the current releases fixes it by configuration: InTouch has no dual-Historian target, the capability Application Server offers. The design that works logs process data in parallel — the same tag list imported into both Historians, with an IDAS on each node reading its own OI Server — and treats historical alarms as a separate stream through A2ALMDB. The checks below run from what the operator sees back to the logging path, then give the build steps, the tag-count arithmetic against the bundled Historian license, and the licensing position for the pair.
What the empty partner Historian looks like #
The partner Historian shows no new rows while the primary fills normally; trends from the secondary node come back empty or stop at the moment the partner relationship was built. Nothing is faulted — no tag error, no driver error — because the data never left the InTouch runtime. Match what you see against this list:
- Primary trends normally, partner has no rows: the built-in InTouch-to-Historian feed targets a single Historian, so build the parallel design below.
- Process data appears on both after the change, alarm history on one only: the alarm path was not duplicated — add A2ALMDB on each node.
- Gaps in the primary Historian whenever InTouch restarts: the built-in feed depends on the InTouch runtime to push values — move acquisition to IDAS reading the OI Server.
- The alarm client reports historical data unavailable after a Historian stop: expected; the client switches to the partner only after the operator acknowledges the dialog, in the version tested.
InTouch has no dual-Historian target #
This is the root cause, not a fault. Application Server writes to two Historians; InTouch does not, and the partner pairing does not change that. The pairing defines the relationship between two Historian nodes while the InTouch runtime keeps historizing to the one Historian named in its application. The capability is under consideration for a future release, and reports point to InTouch 2026, but a roadmap statement is not a setting you can configure. Verify it the direct way: read the release notes and the Historian integration section for the exact version you run. If no secondary Historian setting is listed there — the case for every release shipping today — build the parallel design. If your release does list one, configure it per the documentation and run the failover test below before trusting it.
Feed each Historian through IDAS, not InTouch #
Two acquisition paths exist and they fail differently:
| Path | Where the data comes from | What happens when InTouch stops |
|---|---|---|
| InTouch-to-Historian, the built-in connection | The InTouch runtime pushes tag values | Logging stops for the outage; the gap stays |
| IDAS reading the OI Server | The Historian acquires from the I/O server directly | Acquisition continues without InTouch |
Use IDAS on each node. Field experience from plants running this configuration reports the OI Server feed as the more reliable of the two — a reported result, not a specification — and it matches the failure mode in the table: the history record should not depend on whether a view node is running.
Each Historian then owns its acquisition, the same way you already handle I/O: with two professional communication driver licenses you add a secondary source per device, and the same per-node ownership applies to the historian tier. The cost is discipline — the tag list now lives in three places, InTouch and both Historians, and must stay identical.
Historical alarms need their own path #
IDAS failover carries process data but not historical alarm data into the blocks, so alarm history needs the A2ALMDB alarm-logging solution on each node. Pairing guidance for redundant historians also requires both historians to store events to the same location, so pick that location once and configure both nodes to it. After an incident the alarm record is what the investigation reads, and a single-sided alarm store is the gap people find too late. Verify the stream by raising a test alarm and querying history on both nodes; if one returns nothing, the alarm path is single-sided.
What the alarm client does on failover #
A field test reported for version 2023R2SP1 P02 shows the behavior to plan around: primary and secondary RDSH servers, each with a Historian, configured as partners; two RPP clients logging alarms and historical data into the Historian; clients normally connected to the primary. Stopping the Historian services on the primary made the alarm client (EAC) report that historical data is not available; after the operator acknowledged the dialog, the EAC switched to alarms from the secondary RDSH. The switch is operator-acknowledged, not silent — write that click into the operating procedure. This is a tested observation, not a documented supported feature, so repeat it on your own version before you rely on it.
Building the parallel-logging configuration #
- Pick the client architecture first, because it places the tag source, not the historian design: InTouch Web serving browser clients over HTTPS through the Web Server, or an InTouch tag-server application on a Terminal Server with RDP clients using remote tag referencing. The Historian steps are the same either way.
- Export the tag list from the InTouch application.
- Import or create the same tags in the first Historian and again in the second — the Tag Importer Wizard (IDAS) does the import — with names, engineering units and storage settings identical.
- Configure an IDAS on each Historian node, each reading its own OI Server, and add the secondary I/O source with the second communication driver license.
- Configure A2ALMDB on each node for alarm history, both storing events to the same location.
- Configure the partner relationship between the two Historians.
- Stop the Historian services on the primary, confirm trends and alarms resolve from the secondary, restart, and confirm both nodes resume logging.
If the first Historian already holds history that must exist on the second, treat that as a migration and backfill trend history from a plant historian onto the new node. When a tag resolves on one Historian and not the other after any change, trace the PLC tag from controller to historian display to find which leg dropped.
Checking the bundled 5,000-tag limit for a 4,200-tag application #
Inputs:
- Bundled Historian capacity: 5,000 tags per instance, from the InTouch Unlimited Professional package, one per node.
- Instances in the pair: 2.
- Application size: 4,200 tags — an illustrative input for this check; substitute the count from your own export.
- InTouch tag license: unlimited, so the InTouch side never constrains the count.
Rule in words: parallel logging writes each application tag once per Historian instance, so the per-instance tag count equals the application tag count, and the bundled license limits that per-instance count.
Arithmetic:
- Per node: 5,000 tags − 4,200 tags = 800 tags of headroom.
- Load: 4,200 tags ÷ 5,000 tags = 0.84, so each Historian runs at 84 percent of its bundled license.
- Across the pair: 2 × 4,200 tags = 8,400 stored histories, but licensing counts per instance, so 8,400 never lands in one license.
Result: both Historians fit inside their bundled licenses with 800 tags to spare per node. Decision: below 5,000 tags, the two bundled Historians cover the redundant design with no further licensing; at 5,000 tags or above, license the larger Historian tag count on both nodes before going live, because the limit is per instance and the design duplicates the full list.
Licensing the redundant pair #
Two InTouch Unlimited Professional packages duplicate everything per node: unlimited tags, the bundled 5,000-tag Historian, the professional communication drivers, the web workspaces, reports, and the MQTT and OPC UA servers. The two Historians are separate licensed instances, and the bundled tag count applies to each one. Decision: before ordering, confirm with your distributor — in writing, against the part numbers on the quote — that the Historian license appears on both servers; the redundancy exists only if both nodes carry it.
Questions asked
What does the Partner Historian setting actually do?
It defines the pairing between two Historian nodes, and it is configured in AVEVA Integration Studio. It does not give the InTouch runtime a second write target: InTouch still historizes to the one Historian named in its application. That is why the partner stays empty until you duplicate the tag list and the acquisition on both nodes.
Will moving to InTouch 2026 give me a redundant historian?
Reports point to a redundant historian capability arriving in an InTouch 2026 release, but a roadmap statement is not a setting you can configure today. Read the release notes for the version you run. The parallel IDAS design keeps working after an upgrade, so building it now costs you nothing if the feature ships later.
How do I test the redundant pair before trusting it?
Stop the Historian services on the primary node during a quiet period. Confirm trends still resolve from the secondary, and expect the alarm client to show a historical-data dialog the operator must acknowledge before it switches over. Restart the primary and confirm logging resumes on both nodes before you call the pair production-ready.
How do I keep the tag lists identical over time?
The InTouch export is the master list. Any tag added, renamed or deleted in the application has to be re-imported into both Historians, or that tag has history on one node and a silent gap on the other. Run the import on both nodes in the same change window so the pair never diverges.
Does the bundled 5,000-tag Historian count once or twice?
It counts per instance. Two bundled 5,000-tag Historians store the same tag list twice, which gives you redundancy, not 10,000 distinct tags. The InTouch side is unlimited tags, so the constraint is always the per-instance Historian count on each node.