RequestTimeTooSkewed when retrying transfer of large file

[stdout]

I'm running into an issue that I'm wondering if any one else has
encountered. If during a transfer of a file into S3, a 500 server
error occurs and is caught in connection.py _mexe(), and the failure
occurs more than 15 minutes into the transfer, the first retry always
immediately fails with S3 throwing a "RequestTimeTooSkewed" and the
whole transfer bombs out. The issue is hard to recreate, so I'm sort
of guessing about the "15 minutes into a transfer part", just assuming
that anything > MaxAllowedSkewMilliseconds is when the issue occurs.
Empirically I'm only seeing this with large files that presumably take
longer than 15 minutes to upload.

My workaround is to catch the S3ResponseError and retransmit the whole
file, but that of course wastes a lot of time and bandwidth.

Thanks in advance for any insight anyone has to offer.

[mitch]

I've never actually seen that happen but your explanation seems
plausible. My question would be what's causing the retry in the first
place. In other words, there must have been some other exception
thrown that caused a retry of the request in the _mexe code. In fact,
there were probably multiple retries and eventually one of the retries
caused the "RequestTimeTooSkewed" response which is unrecoverable.

There is logging in there that would tell you about the initial errors
and retries but only if you have your log level set to DEBUG. You
could enable that if you feel there is a reasonable likelihood that
you will encounter the error again.

Mitch

[stdout]

Sorry for the delay in responding. Before your reply, I decided to
work around the issue by breaking files up in to smaller chunks that
would transmit in < 15 minutes, and have not seen the issue since.
Thanks for the great library BTW.