Import FAQ

My import looks stuck — the count stopped moving

Usually normal, for one of a few reasons:

• It is waiting its turn. Only one archive unpacks at a time, only one reads at a time, and only one saves at a time. When you import a batch, the later archives show Waiting to start for as long as the earlier ones take. A batch of thirteen archives can leave the last one waiting for close to an hour.

• A single large file is being processed. One long video or one enormous mailbox can hold up the rest of a batch. The count moves per record, so a record that takes a minute looks like a stall.

• Disk buffers are flushing. Writing a lot of data looks fast while it fills an operating-system buffer, then pauses while the buffer drains to the device. Watching the machine’s disk-write metric confirms it. Faster storage shortens these pauses.

• The source has to be read in order. Some exports interleave data from several devices to keep it chronological — the classic Google Location History does — so the parser has to read ahead to find the next record it wants.

• The stage has no denominator yet. If the bar is animating with a count but no percentage, the parser has not yet worked out how much there is. That is not a stall; a percentage would be a guess.

• Its worker stopped. If the line under the import reads Stalled — no heartbeat, the stage has reported nothing for 90 seconds. When a worker died partway — a restart in the middle of a read, say — Companion Memory takes the stage back and retries it by itself within about four minutes, and the line then reads “retrying, its worker stopped”. If it is still stalled after that, open the run in History and use Try again.

If a stage shows Failed, it is not stuck: open it in History and use Try again.

Why did my import finish as “partial”?

Because some of it did not import, and saying so is more useful than either claiming success or reporting failure. The status reads Imported, with some items skipped.

The common causes are one parser failing while others succeeded, an archive in a multi-part export being unreadable, or files or records that the Store files or Save stage could not write. Where the run can say which, it lists them under the stage that noticed. Skipped groupings such as albums do not make a run partial; see Why are my photo albums missing?.

Can I import the same archive twice?

Yes, and it will not duplicate anything. Every record carries where it came from and its identity within that source, so a second import recognises what is already there.

This is also what makes re-importing after an upgrade worthwhile: improvements to a parser reach records you already imported.

Do I need to unzip my export first?

No. Upload the .zip as it downloaded. The appliance expands it, including archives nested inside it, which is how Google Takeout ships.

What happens to my uploaded file?

Your browser sends a copy, and your original stays where it is. The appliance stores the copy while the import runs, expands it into a working directory that belongs to that run, and deletes the archive as soon as it has expanded and the working directory when the import finishes. An upload that you never import is deleted after about a day.

The appliance does not keep your archive for you. Keep your own copy.

If you copy files into the inbox folder instead, the appliance treats them as copies and deletes them as their import consumes them — copy, do not move, your only copy.

How long does the appliance keep an import’s data?

While an import runs, the appliance holds a staged copy of every parsed record. It deletes the staged copy of a completed import after seven days, and keeps the run’s entry in History for 90 days. A read import that you never saved is kept until you select Sync to my memory or Discard, because the staged records are the only copy of the parse.

Does any of my data leave the appliance?

The archive is stored, expanded and read on the appliance, and the records it produces are written there.

The exception is exports that reference media by link rather than including it, such as Reddit’s. Companion Memory saves the link and not the file, and your browser loads the media from the service that hosts it when you view the record.

After the import, two background jobs can send limited data out: reverse geocoding sends place names and coordinates to CI Portal, which relays them to a geocoding provider, and link previews are fetched from the sites your links point to. See What leaves your appliance.

Why is one person showing up twice?

People are matched on a normalised handle — a lower-cased email address, or the last ten digits of a phone number. Someone who appears under two different addresses is two people as far as the importer can tell, because nothing in the export says they are the same.

Importing a contacts export helps, since it is the one source that states which addresses belong to whom. For the ones that remain, Memory → People has a merge-suggestion queue you can accept or dismiss.

Why are my photo albums missing?

Album, thread and place groupings are skipped at present. They have no stable identifier the way a person has an email address, so importing them repeatedly would create a new copy each time rather than recognising the existing one.

The photos themselves import normally. It is the album grouping that is not yet carried across. The run reports how many were skipped: “N groups skipped (albums, threads and places)”.

Why does my contacts import say “0 records”?

The records count on a finished import counts items saved as messages, events, documents and the like. A contact becomes a person, not a record, so an import of only contacts reports 0 records even when it worked. Check Memory → People to see them.

Can I import only part of an archive?

Not selectively. Every archive imports whole. To leave something out, remove it from the export before you upload it — for example, drop the unzipped folder without the subfolders you do not want.

What does “Discard” do?

Discard, on the Sync step, deletes the parsed rows staged for that import. It does not delete your export: if the appliance still holds the archive you uploaded, it goes back to your uploads. Photos the import already stored stay in your memory. Records already saved into your memory are ordinary records and are not removed.

How do I remove data an import already saved?

There is no button that removes everything one import saved. You can delete records one at a time from the Explorer’s detail drawer, delete a connected service’s data by deleting the connection, or delete your account under Self → Profile.