Explainer
Where the Ignition Edge tag historian writes its data
Short answer
Ignition Edge has no configurable path for its local tag historian: the store and its store-and-forward buffer both write to the volume Ignition is installed on. For a 7-day history on fewer than 100 tags, add 3 to 5 GB of headroom to the main partition; an external SQL database connection is the only supported way to choose the volume.
On this page
Ignition Edge gives you no setting that chooses the partition or volume for its local tag historian. The built-in storage provider writes on the volume where Ignition is installed, and the store-and-forward buffer that feeds it lands on the same volume, so the only supported way to pick the volume is to point the historian at an external SQL connection and put that database engine where you want it. For the common case, a 7-day history on fewer than 100 tags, neither store is large: give the main partition 3 to 5 GB of headroom above the minimum and image that, rather than moving the historian file onto a secondary partition with a link.
What the local historian actually is #
Ignition Edge ships with an internal storage provider for tag history, so a historian runs on the gateway with no external database at all. Its page on the gateway is Config > Gateway Settings > Historian. In the 8.1 documentation, every historical value passes through the store-and-forward engine before it is written to the storage provider the historian is tied to, and the provider keeps the metadata that travels with the values — tag path, datatype, units — in system-managed tables. Those tables are the ones behind why Ignition tag historian tables are not created when a connection is fresh.
The local store is bounded by design: the documented ceiling is 35 days and 10 million points, and pruning holds the store to that boundary. If you ever run more than one historian connection, why two Ignition historian connections share one pruning boundary explains how their retentions interact.
Where each write lands #
| What writes | Where it lands | Can you choose the volume |
|---|---|---|
| History on an external SQL connection | Wherever that database engine keeps its files | Yes — the database connection targets the instance |
| The Edge local storage provider | The volume Ignition is installed on | No — no path setting exists; a link after the fact is the only workaround |
| The store-and-forward buffer | The volume Ignition is installed on | No — it lands on the install volume in every configuration |
The gateway exposes no path field for the local store, and no conf-style argument for it exists today. If you need one, that is a feature request to put to the vendor, not a setting you have missed. Before changing any layout, open the tag history configuration and read the storage provider it names. If it names a SQL connection, history already goes wherever that engine keeps its files, the partition work below does not apply, and you size that server instead — evaluating an Ignition historian on TimescaleDB walks through that class of decision.
Sizing the main partition for a 7-day history on 100 tags #
Inputs:
- Tags historized: fewer than 100 on this gateway, so 100 is the bounding count.
- Retention: 7 days.
- Documented ceiling of the local store: 35 days and 10,000,000 points.
- Drive to be imaged: 128 GB.
Rule in words: the ceiling bounds both the window and the rate. A 7-day window can hold at most 7/35 of the point ceiling, and the ceiling fixes the highest sustained rate the store can write.
- 10,000,000 points × (7 ÷ 35) = 2,000,000 points in the window.
- 10,000,000 points ÷ 35 days = 285,714 points per day.
- 285,714 points per day ÷ 86,400 s per day = 3.3 points per second across all tags.
- 3.3 points per second ÷ 100 tags = 0.033 values per tag per second, about one stored value per tag every 30 s.
At that rate and tag count a 7-day file stays under 1 GB, and it cannot exceed 2,000,000 points whatever the tags do. The store-and-forward buffer shares the same volume.
Decision: add 5 GB of headroom to the minimum partition holding the OS and Ignition; 3 GB is the floor when the layout is tight. After the first full 7-day window, read the file size on disk. Under 1 GB confirms the sizing; above it, grow the partition before you take the image.
Moving the file with a link, and when to accept it #
Accept this only when the main partition cannot grow. The file has to exist first, so run these steps after the historian has been writing.
Identify the file by observation rather than assumption. With tags logging, list files under the Ignition install directory by last-write time, twice, a couple of minutes apart. The historian file is the one whose timestamp keeps advancing; if data, journal and lock files change together, treat the containing directory as the unit to move.
Windows (PowerShell):
$ign = "C:\Path\To\Ignition"
Get-ChildItem -Recurse -File $ign | Sort-Object LastWriteTime -Descending |
Select-Object -First 15 FullName, Length, LastWriteTime
Linux:
ign="/path/to/ignition"
find "$ign" -type f -mmin -5 -printf '%TT %s %p\n' | sort -r
- Stop the Ignition service. Copying a database file mid-write yields an inconsistent copy, and the gateway keeps writing to the open original — this is the one step not to bend.
- Copy the file or directory to the secondary partition and confirm the copy matches the original size.
- Rename the original with a
.baksuffix. - Create the link at the original path:
mklink "C:\Path\To\Ignition\file" "D:\History\file" # Windows, file link
mklink /D "C:\Path\To\Ignition\dir" "D:\History\dir" # Windows, directory link
ln -s /secondary/history/file /path/to/ignition/file # Linux
- Grant the account that runs the Ignition service read and write access to the target folder.
- Start the service, then confirm the last-write time advances at the new path while the
.bakstays frozen. If it does not, the gateway is writing somewhere you did not link: stop the service, remove the link, restore the original, and repeat the identification step.
A drive swapped from an image made this way carries a link whose target lives elsewhere, so three conditions must hold on the replacement drive:
- The secondary partition mounts at boot at the same path or drive letter, set through Windows volume assignment or the Linux mount table.
- The target folder exists on that partition; create it as part of the swap procedure.
- The service account has write access to the target.
Miss the first or second and the link dangles with no valid file location; miss the third and the gateway runs without writing history.
Keeping the swap image small #
Two routes work on the example drive. A whole-drive image produces a raw 128 GB file that compresses to roughly 8 GB because most of the drive is empty — a long read and a large temporary file, but no layout changes. A sized main partition images in a file barely larger than the partition itself, with no link or mount dependency inside it.
Decision: use the sized-partition image unless your tooling only takes whole drives. History does not need to be inside a recovery image; it accumulates again after the swap.
Checking the layout before you rely on it #
- Query a recent range on a historized tag in a trend and confirm the data is continuous across the service restart — no gap where the copy happened.
- After one full retention window, read free space on the main partition. The headroom has to cover the store-and-forward buffer as well as the historian file.
- If you moved the file, rehearse the failure once with the secondary partition unmounted: start the gateway and read the status pages and logs, so a failed swap is a condition you have already seen.
Questions asked
Does the store-and-forward buffer move with the historian file?
No. Store and forward is a gateway feature: its buffer is written on the volume where Ignition is installed, whichever historian the tags target. Even with history on an external SQL server, the buffer stays on the gateway's install volume, so the main partition needs room for it in every layout.
How much can the Edge local store hold?
The documented ceiling is 35 days and 10 million points, and pruning enforces it. That ceiling, not the disk, is what stops a runaway store: even a gateway logging at the maximum fills to 10 million points and then old data is pruned, so a 7-day window at sub-100-tag rates never approaches it.
Can I copy the historian file while the gateway is running?
No. The copy can capture the file mid-write while the gateway keeps writing to the open original, leaving the copy stale or corrupt. Stop the Ignition service before the copy and start it again only after the link exists and the target permissions are in place.
What has to be true on the replacement drive after a symlink move?
Three things: the secondary partition must mount at boot at the same path or drive letter, the target folder must exist on it, and the service account must have write access to it. Miss any one and the link dangles, or the gateway runs without writing history.
How do I confirm whether my tags use the local store or an external database?
Open the tag history configuration and read the storage provider it names, and check the historian settings page on the gateway. If a SQL connection is named, history goes wherever that database engine keeps its files, and the gateway partition only needs room for the buffer.