Closed m-petersen closed 3 years ago
thanks for including WTF. so it is happening at git-annex level...
chunk=1GB
to your initremote
call . That should chunk up files over 1GB into separate subkeys on the remoteencryption=shared
so that key will be stored in git-annex repo (so not avail on S3) thus even if someone gets access to S3, data will be safe as long as they don't have access to dataset (having recommended that: I personally and overall throughout datalad AFAIK we didn't test such feature of git-annex special remotes extensively, but I expect it to work)<details>
<summary></summary>
</details>
edit: FWIW I use chunk=1GB
for backing up to dropbox via rclone, with many files way over 1GB limit -- never had an issue.
@m-petersen could you please confirm that you cannot upload files over 5GB to s3 remote without chunking with a fresh git-annex?
Sorry for the delay.
Just tested your instructions and using chunk=1GB
when initiating the special remote resolved the issue.
I also appreciate your recommendation considering the encryption.
What is the problem?
I aim for establishing a S3 special remote for our local S3 Bucket in a dataset of singularity containers to make that dataset shareable across clients. When datalad tries to upload container images >5gb it fails with
I contacted the sysadmin for our S3 storage and he said that the Bucket is configured to allow files >5gb. His take on this is that the S3 client used by datalad under the hood won't switch upload "strategies" as for example aws cli would do (PUT for objects <5gb and MPU for >5gb).
The full verbose output is
``` $ datalad -l debug push --to github [DEBUG ] Command line args 1st pass for DataLad 0.14.4. Parsed: Namespace() Unparsed: ['push', '--to', 'github'] [DEBUG ] Discovering plugins [DEBUG ] Building doc forWhat steps will reproduce the problem?
I initialized the S3 special remote as descriped in a handbook chapter (https://handbook.datalad.org/en/latest/basics/101-139-s3.html) with a script.
What version of DataLad are you using (run
datalad --version
)? On what operating system (consider runningdatalad wtf
)?WTF
# WTF ## configurationAs always grateful for any input!
Cheers, Marvin