Say a 400+MB dataset. The download will fail at like 80MB. Doesn't matter whether using SDK or from clearML experiment page.
Hmm BitterStarfish58 what's the error you are getting ?
Any chance you are over the free tier quota ?
Error message as the image says.
The file size is 415MB, but the download "succeeds" at 107MB.
BitterStarfish58 I would suspect the upload was corrupted (I think this is the discrepancy between the files size logged, to the actual file size uploaded)
AgitatedDove14 In our case, redownload doesn't help because it leads to the same result. The download gets interrupted due to network error.
But it's a good start to tell finishing and succeeding apart
Since the error says network error, is it possible because I'm in Taiwan? Like downloading from Asia leads to this kind of issue.
It shows we're still in the free tier quota
https://clearml.slack.com/archives/CTK20V944/p1642735039222200?thread_ts=1642731461.221700&cid=CTK20V944
Like I said here, using browser doesn't work. It has the same behavior.
Since the error says network error, is it possible because I'm in Taiwan? Like downloading from Asia leads to this kind of issue
Can you download it from the browser ? (I mean the file size after download , is it 400mb?)
AgitatedDove14 Earlier my colleague said he managed to download the dataset with browser by keeping "resuming" the download once it stops due to network error. So no I don't think it's the problem of the file itself...
Regrading check, we can add one here
https://github.com/allegroai/clearml/blob/12fa7c92aaf8770d770c8ed05094e924b9099c16/clearml/storage/helper.py#L713
I'm not sure the files-server supports "continue" from last position...
AgitatedDove14
https://github.com/allegroai/clearml/issues/552
Just did. Hope the format looks okay.
Hi BitterStarfish58
Where are you uploading it to?
Pretty sure it's not the reason. Now I've encountered this issue on 5+ dataset I'm using on different projects. Some worked quite well before, but not recently.
So are you saying the large file size download is the issue ? (i.e. network issues)
Hmm maybe we should add a test once the download is done, comparing the expected file size and the actual file size, and if they are different we should redownload ?
I'm not familiar with that either. But downloading with chrome browser and some perseverance to keep clicking continue does work. It's quite cool.
(currently I think the implementation expects that if the download completed, it was successful)
BitterStarfish58 could you open a GitHub issue on it? I really want to make sure we support it (and I think it should not be very difficult)
Anyway, thanks for the help. As a workaround, we will avoid large file uploading from now on. Look forward to hearing from you if you guys manage to reproduce the issue or implement a fix.
AgitatedDove14 Yes I think that's the problem. And if there's also a way to keep resuming the download when using sdk, our python code will work like before. That's basically all we need.