Summary
I’m unable to download a full, non-corrupted exported archive of my cloud instance. Using mmctl, the download gets interrupted randomly and there is no way to recover or continue the download.
Steps to reproduce
I followed the official guide to export the data: Mattermost workspace migration — Mattermost documentation
- mmctl login
- create an export archive
- mmctl export download
Expected behavior
The entire archive gets downloaded in one session and, if not, it would be resumed at the next attempt.
Observed behavior
- There are no error messages, no visible retries or multiple attempts: the server closes the connection while downloading, seemingly at random, and the archive gets truncated.
- If I try to download the file multiple times, every time I get an archive of a different size.
- If importing the archive is attempted, the import will fail because the archive is corrupted.
- I am currently trying to work around this issue via a custom script: I’m calling
mmctl export download
in an infinite loop, discarding a download if it is smaller than the previously downloaded file and it’s been running for hours. - The cloud mattermost server does not include
Content-Length
in the response header, making impossible to know in advance how large is the file to be downloaded. - The
mmctl
client is unable to resume a broken download. - I dived a bit into the code, but this issue seems to be server-side, so there’s not much I can do: mattermost cloud is managed by the mattermost team, so I’m locked in until this issue is fixed or I’m lucky enough that a non-corrupted archive is downloaded.