Perl dies with error message during backup up using S3's 'S3-compatible server hostname' option to backup to Google Cloud Storage instead of Amazon.

The Perl script above is wrong it has the wrong server and port. Correct code is below:

use strict; use warnings; use IO::Socket; use Data::Dumper;

my $sock = new IO::Socket::INET (PeerAddr => 'storage.googleapis.com', PeerPort => 443, Proto => 'tcp'); die "$!" unless $sock;

print Dumper($sock);

Thanks, Dave. Perl script output: [root@web5 ~]# perl test.pl $VAR1 = bless( *Symbol::GEN0, 'IO::Socket::INET' );

As mentioned I'm currently using Virtualmin GPL, and the google cloud storage support is a premium feature only currently, so I suppose that's the only reason why I'm not using it. Virtualmin seems to support other cloud's S3 compatibility with this option, so I though I'd give it a shot.

I decided to report this as a bug instead of a forum post, because I was getting a perl error message.

First I used telnet, and it worked fine except it disconnected, because it expects SSL stuff to happen not http right away, because its https, and the ssl stuff needs to happen.

Below I'm using openssl's S_client, which is like a telnet that supports ssl, and it manages to get an error message from google's api servers. So, I'm pretty confident that there is no connection, networking, or firewall problem between my server, and google's.

I think there's a real bug here somewhere, but it's strange, because I'm sure loads of people are using the S3 backup option, so maybe the problem is perhaps just in the code where the different hostname is used?

Google's server is telling me my request is no good, so this means I can connect to storage.googleapis.com:443 just fine. I think there might be a real bug here.

Please try my steps to reproduce the bug. It's very easy to reproduce as it happens every time.

Thanks.

Here are faster and easier steps to reproduce this problem:

1. Change the hostname from the default amazon one to google's "storage.googleapis.com" in Backup and Restore -> Cloud Storage Providers -> Amazon S3 -> S3-compatible server hostname should be set to "storage.googleapis.com". Don't forget to change the radio button to the one next to the text box!
2. Now just go to Backup and Restore -> Amazon S3 Buckets
3. Below is the error message you get:

Try this way of reproducing the problem as it's faster, and you won't have to set up a backup; however, source code file and line number are different than the one used for backup, so both places or perhaps just the same function both places call will need fixing.

Thanks.

This could actually be the cause of the problem.

Our server does have a IPv6 configuration, but it's quite a few years old before IPv6 was really needed, so our server has a IPv6 configuration for our host's internal network, but it has no publicly available IPv6 IPs that can reach the internet.

If Virtualmin is setup to prefer IPv6 if IPv6 is available, then this could be the cause. I don't know if it makes sense to add support for dropping down to IPv4 if IPv6 gives a network unavailable error, or if supporting that makes no sense.

I've created a support ticket with our host to get them to add a public IPv6 address to our server. Once that's done, if IPv6 being privately configured, but not publicly configured is the cause, then that should fix this problem.

I'll update this ticket after the server gets a IPv6 address, and I retest it.

Thanks.

Thanks, for figuring out the IPv6 problem. It turns out the server already had a global internet usable IPv6 address allocated, but for whatever reason it was never configured on the server.

I configured it, and backup works great. It uploads to the amazon S3 bucket just fine!

But it only works when you specify a specific S3 bucket that has already been created. If you use the %d-%m-%Y option and the create destination directory option and the delete after 1 day option it fails.

Is this option supported in S3? It seems like it should work just fine.

When you use the "Do strftime-style time substitutions on file or directory name" you get the error message below.

Also when you use the Amazon S3 Buckets page the buckets are listed just fine, but when you click on one you get this weird Perl syntax error:

IPv6 works great now, but still getting weird Perl errors when using the strftime-style backup option.

To reproduce configure google's S3 as the endpoint as described above, then try listing S3 buckets' contents in the Amazon S3 buckets page, and then configure a S3 backup using the strftime-style backup option with the create directory option set too and the delete after 1 day option too. Then just run the backup manually, and you should see the error message.

Please look into these weird errors.

Thanks.