I tried to replicate and migrate the openclaw service to a Hetzner server, but the replication failed. Could you please help me identify what went wrong?
Looking forward to your assistance, thanks.
ForumForumI tried to replicate and migrate the openclaw service to a Hetzner server, but the replication failed. Could you please help me identify what went wrong?
Looking forward to your assistance, thanks.
Evicted: The node had condition: [MemoryPressure].
It looks like the issue was caused by the backup file being too large. We are working on a fix, sorry for the inconvenience! In the meantime, could you please restart your server to free up the memory occupied by this?
I restarted it, thanks!
After restarting the server, I tried copying it again, but it still failed.
The underlying issue has not been fixed yet. It is recommended to use the backup + restore method for now: https://zeabur.com/docs/zh-TW/operations/data/backup-restore
I don't understand, didn't you ask me to restart the server earlier?
I just finished backing up, do I still need to restart it?
When can it be copied to the new server? Otherwise, I can't use it on either side.
Hello,
We have re-checked your migration status, and we have good news:
The data has been successfully copied to the Hetzner server. Your brandclaw-meront project on Hetzner now has 502MB of OpenClaw data (including the openclaw.json configuration file and related data).
However, the current issue is: the service is not running — there are no active Deployments in the brandclaw-meront project on Hetzner.
Please try the following:
The OpenClaw instance on your original server (BrandClaw, Tencent Tokyo) is currently stopped as well. The data is now on Hetzner, so it should work normally once started.
Please let us know the result.
Additionally, if you are unable to open it, please let me know the reason!
Hi, based on the data structure I see, it appears to be the result of a failed migration (as shown in the failed status screenshot I sent earlier).
Although openclaw starts up normally, none of the data from my old server is present.
Please help me investigate what is causing the project cloning/migration issue. Thank you!
Hello, we have re-examined your migration status and found the cause:
Issue: Only part of the data was copied. Your OpenClaw data on the original Tencent Tokyo server totaled 4.4 GB, but only 502 MB (approx. 11%) was copied during the migration to Hetzner before it was interrupted due to an Out of Memory (OOM) error on the source server. Therefore, what you see on Hetzner is an incomplete dataset, which is why OpenClaw does not have your old data after starting.
Your original 4.4 GB of data remains intact on the Tencent Tokyo server and has not been lost.
Recommended Solution:
Since Zeabur's automatic migration feature may fail when the source server runs out of memory, we recommend using manual backup and restore instead:
This will bypass the memory limitations of the automatic migration and move the complete data directly.
If you encounter any issues during the backup/restore process, please let us know and we will assist you.
Alternatively, you can use the built-in backup feature of openclaw and restore it on your Hetzner machine.
I tried to upload the backup file "backup-*.zip" (from Zeabur's built-in backup feature) to the /home/node folder, but I received the error: "upload failed: The requested file could not be read, typically due to permission problems that have occurred after a reference to a file was acquired." How can I resolve this?
One more thing, will the 'Clone Project' feature be fixed in the future?
I used the clone project feature because I needed to upgrade the server due to insufficient memory. I look forward to Zeabur making it easier for users to migrate and upgrade!
Hello,
Regarding the upload failure error:
The error message "The requested file could not be read, typically due to permission problems that have occurred after a reference to a file was acquired" is a browser-side error, not a permission issue on the Zeabur server. Common causes include:
We recommend trying these steps first:
backup-*.zip file to a simple local folder like "Downloads" or the Desktop, and verify in Finder that the file icon does not have a cloud download symbol (☁️).Alternatively, skip the web upload and use the Command tab to restore directly:
If the upload continues to fail, you can download the backup file from the Files tab of your old project (Tencent Tokyo) OpenClaw service, then use OpenClaw's own backup/restore process. Or, run this directly in the Command tab of the OpenClaw service on the new server:
restore backup-xxxx.zip --strip 2
(--strip 2 applies to files generated by the Zeabur backup service). Restart the service after execution.
Regarding the "Copy Project" feature:
Our engineering team is aware of this issue. It is indeed caused by the source server running out of memory while copying a large amount of data (4.4 GB in your case), leading to an interruption. We will work on improving this process to make future upgrades and migrations smoother. There is no specific fix time yet, so we recommend using the backup and restore method to handle your migration needs for now.
If there is any progress or if you get stuck during the restore process, please reply and we will continue to assist you.
Hi, although I haven't encountered these issues when uploading before, I tried uploading again and it still doesn't work...
My steps followed the official Zeabur documentation for backup services: https://zeabur.com/zh-TW/templates/VTZ4FX#備份與還原
/home/node folder.backup*.zip directly from Finder to the node folder.It shows the same error.
Note: I also tried uploading from my desktop, but it resulted in the same error.
I performed another test: I uploaded a 1MB image file from my downloads folder, and it uploaded successfully.
My guess: Is the zip file too large?
Could you please help me with the migration directly?
It has been over a week now... I'm not sure when the migration will be successful, and both servers are currently unusable...
Please perform a backup of the current status again, and we will assist in handling it.
Just ran the backup, thanks!
We have assisted in uploading the files, performing the restore, and restarting the service.
Next time you need to do this, please follow these steps to upload the files:
npm cache clean.restore <filename>, then wait for the file to be extracted and overwritten.Thanks for the help!
Just to clarify:
Regarding your first point, you mean that for an old service, only one Zeabur backup file can exist, right?
Is this related to why I cannot upload files within the Zeabur service on the new server?
Also, regarding your third point, how can I download a Zeabur file using Google Drive?
The first point is mainly about the issue of having other backup files within the backup. Since Zeabur's backup is a full disk backup, if it contains other backup files or temporary files, the file size will become too large (perhaps compressed files packed by Openclaw's own backup function?).
Currently, files that are several GBs in size cannot be uploaded via the web interface, so other methods must be used to place them on the server, such as third-party storage or scp.
For the third point, you can ask AI which cloud storage services support large file uploads and downloads. However, I found that using Google Drive for files over 100MB is very troublesome and not recommended.
It is mainly done by using the curl command to download files to the disk.
I see, I guess I'll wait until the "Duplicate Project" feature is fixed. 🤣
Will there be an announcement on Discord once it's fixed?
It's not actually broken; it just doesn't support enough scenarios yet. Specifically, the project cloning feature doesn't support large files, which is a requirement for openclaw, and we haven't implemented that yet.
You can try using git to save your workspace settings and install packages separately. https://docs.openclaw.ai/concepts/agent-workspace#git-backup-recommended-private
You mentioned previously that the development team was aware of this. Will this feature be prioritized?
We are currently planning a deployment solution for our clients. If we need to upgrade, it would be much more convenient if we could directly clone projects to a new server.
If it's not possible to migrate a 2GB project, we might have to reconsider for now.
Backup and restore (in development) are features we value, but there is a more significant update this week, so it might be slightly delayed.
This post has been inactive for a while. We will be closing it in 2 days if there is no new activity.
New replies are disabled for resolved issues.