You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When using an HTTP (not an HTTPS) proxy, we do not handle the case where the proxy server will send headers along with the 100 continue response when we set the Expect header in our request. This only happens for specific S3 operations, but when using these S3 commands, the AWS CLI will hang and eventually exit with a timeout exception:
Expect: 100-continue (same)
CLI -> SQUID -> S3
HTTP/1.1 100 Continue HTTP/1.1 100 Continue
Connection: keep-alive
CLI <- SQUID <- S3
object body object body
CLI -> SQUID -> S3
200 OK 200 OK
CLI <- SQUID <- S3
Note how from SQUID to CLI a Connection: keep-alive is injected in the proxies response.
The text was updated successfully, but these errors were encountered:
Most servers (including S3) will just return the CLRF after the 100
continue response. However, some servers (I've specifically seen this
for squid when used as a straight HTTP proxy) will also inject a
Connection: keep-alive header. To account for this we'll read until we
read '\r\n', and ignore any headers that come immediately after the 100
continue response.
Fixesaws/aws-cli#1116
When using an HTTP (not an HTTPS) proxy, we do not handle the case where the proxy server will send headers along with the 100 continue response when we set the Expect header in our request. This only happens for specific S3 operations, but when using these S3 commands, the AWS CLI will hang and eventually exit with a timeout exception:
Note how from SQUID to CLI a
Connection: keep-alive
is injected in the proxies response.The text was updated successfully, but these errors were encountered: