[Cyberduck-trac] [Cyberduck] #11416: S3 path-style requests not working

[Cyberduck]

#11416: S3 path-style requests not working
----------------------+----------------------
 Reporter:  georgp    |         Owner:
     Type:  defect    |        Status:  new
 Priority:  normal    |     Milestone:
Component:  s3        |       Version:  7.7.2
 Severity:  normal    |    Resolution:
 Keywords:            |  Architecture:
 Platform:  macOS 11  |
----------------------+----------------------
Description changed by dkocher:

Old description:

> Hi,
>
> according to the changelog of 7.2.0 and this ticket
> (https://trac.cyberduck.io/ticket/10956), the possibility to configure
> path-style for S3 was removed. Later two mechanisms were added to
> automatically detect when path-style should be used:
>
> - Situation 1: When server is an IP address:
> https://trac.cyberduck.io/changeset/48257
> - Situation 2: When A/AAAA record for bucket.hostname.example is answered
> with `NXDOMAIN`: https://trac.cyberduck.io/changeset/48332
>
> However, there are use-cases when both situations don't apply but path-
> style might still be required for the connection to work.
>
> For me situation 1 will not apply as I have to use a domain name so the
> SSL certificate can be verified. Situation 2 also does not apply because
> we're using wildcard DNS records for `*.hostname.example`. In this case
> Cyberduck checks for a A record of `bucket.hostname.example` which
> obviously exists but it does not point to the correct system as the S3
> system is configured to work with path-style buckets only.
>
> Other people are having the same issue with wildcard DNS setup
> (https://trac.cyberduck.io/ticket/10956#comment:10) or with a nameserver
> of their provider which answers to requests with own IP addresses
> (https://trac.cyberduck.io/ticket/10888#comment:31, bad practice but it's
> out there..).
>
> Therefore it would be great to have a config option to disable the
> automatic detection and specifically set the mode to path-style.
>
> Besides that it seems that the mechanism for situation 2 (auto-detect
> with DNS lookup) is broken on MacOS 11.0.1 and Windows 10, Cyberduck
> 7.7.2. I tried it with a different S3 system that does not rely on
> wildcard DNS records. I am able to connect to the system, list buckets
> and files but as soon as I initiate a download, an error message appears,
> clearly indicating it is trying to resolve the non-existing domain and
> fails:
>
> {{{
> Failure to read attributes of file.png.
>
> bucket.hostname.example. DNS is the network service that translates a
> server name to its Internet address. This error is most often caused by
> having no connection to the Internet or a misconfigured network. It can
> also be caused by an unresponsive DNS server or a firewall preventing
> access to the network.
> }}}
>
> So it seems as if the auto-detection works for the initial connect and
> file browser but not for downloading files.
>
> Let me know in case I should provide more details.

New description:

 Hi,

 according to the changelog of 7.2.0 and this ticket (#10956), the
 possibility to configure path-style for S3 was removed. Later two
 mechanisms were added to automatically detect when path-style should be
 used:

 - Situation 1: When server is an IP address:
 https://trac.cyberduck.io/changeset/48257
 - Situation 2: When A/AAAA record for bucket.hostname.example is answered
 with `NXDOMAIN`: r48332

 However, there are use-cases when both situations don't apply but path-
 style might still be required for the connection to work.

 For me situation 1 will not apply as I have to use a domain name so the
 SSL certificate can be verified. Situation 2 also does not apply because
 we're using wildcard DNS records for `*.hostname.example`. In this case
 Cyberduck checks for a A record of `bucket.hostname.example` which
 obviously exists but it does not point to the correct system as the S3
 system is configured to work with path-style buckets only.

 Other people are having the same issue with wildcard DNS setup
 (#10956#comment:10) or with a nameserver of their provider which answers
 to requests with own IP addresses (#10888#comment:31, bad practice but
 it's out there..).

 Therefore it would be great to have a config option to disable the
 automatic detection and specifically set the mode to path-style.

 Besides that it seems that the mechanism for situation 2 (auto-detect with
 DNS lookup) is broken on MacOS 11.0.1 and Windows 10, Cyberduck 7.7.2. I
 tried it with a different S3 system that does not rely on wildcard DNS
 records. I am able to connect to the system, list buckets and files but as
 soon as I initiate a download, an error message appears, clearly
 indicating it is trying to resolve the non-existing domain and fails:

 {{{
 Failure to read attributes of file.png.

 bucket.hostname.example. DNS is the network service that translates a
 server name to its Internet address. This error is most often caused by
 having no connection to the Internet or a misconfigured network. It can
 also be caused by an unresponsive DNS server or a firewall preventing
 access to the network.
 }}}

 So it seems as if the auto-detection works for the initial connect and
 file browser but not for downloading files.

 Let me know in case I should provide more details.

--

--
Ticket URL: <https://trac.cyberduck.io/ticket/11416#comment:1>
Cyberduck <https://cyberduck.io>
Libre FTP, SFTP, WebDAV, S3 & OpenStack Swift browser for Mac and Windows