@JBB said:
Weird stuff with pricing:
g1.nano shows 9/year, Stripe checkout shows 10/year for node S (!)
g1.micro shows 18/year, Stripe shows 18/yr for node M (ok)
g1.small shows 3.3/month, Stripe shows 3/month for node L (!)
If you move the slider to enable annual billing for g1.small and grab the 20% discount, it shows 36/year, but at checkout Stripe shows 32/year. Isn't 0.8 * 3.3 * 12 actually 31.68?!
After clicking through the pricing pages (but not completing the Stripe checkout), I've now got a bunch of pending nodes awaiting payment. Again, didn't expect that. Is there a way to back out of the deal cleanly?
I created a couple of g1.nano nodes, ran out of ram trying to build the spoofer, destroyed them, and created two new g1.small nodes. The IPv6 addresses were re-used, which surprised me!
The goal is to assign a /64 for each node, right?
BTW, is there a NAT64 bridge by default? I noticed some IPv4 only hosts magically worked!
Thank you for your extensive testing!
yea I’m aware of the pricing differences, most plans currently have the wrong price/ plan ID configured so stripe shows the pricing of a different plan.
I’ll create the final plans later and set the proper IDs.
The failed checkout attempts should get cleaned automatically after a few days, but ill add a button to delete them manually.
The goal is to assign a routed /56 to each node, but that might not be possible in each location.
Yes, the nodes have a NAT64/ DNS64 namserver configured by default to allow access to IPv4 only resources like GitHub
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
@JBB said:
Tried the domain feature: with port set to 80 both A and AAAA records created, I'm getting a SSL_ERROR_INTERNAL_ERROR_ALERT from Firefox. curl -v shows there's a 308 Permanent Redirect from http to https, but no cert? (curl is showing error:14077438:SSL routines:SSL23_GET_SERVER_HELLO:tlsv1 alert internal error).
Hmm...maybe I didn't wait long enough for caddy to grab a cert, as it seems to be working now.
Might be slick if the control panel showed the status (active or configuring) so you'd know not to hit it until it was ready.
Yep totally agree, I'll add a info text or find a way to query caddy for the current ssl status.
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
I didn't try your beta, just clicked on the link in your signature, which I guess is the same project.
It doesn't seem possible to determine what regions you provide services in without going through the ordering process, and the first step of that requires you to register. Maybe it's just me, but as a user, I prefer to visit the order page a couple of times checking out the options etc, maybe coming back in a few days, before actually committing and I'd expect registration to be integrated in the last step of the checkout process.
I appreciate that as a business you want to grab an email address as soon as possible, but personally I think that if you ask for it before the user is actually committed to the purchase, you risk putting them off entirely.
@ralf said:
I didn't try your beta, just clicked on the link in your signature, which I guess is the same project.
It doesn't seem possible to determine what regions you provide services in without going through the ordering process, and the first step of that requires you to register. Maybe it's just me, but as a user, I prefer to visit the order page a couple of times checking out the options etc, maybe coming back in a few days, before actually committing and I'd expect registration to be integrated in the last step of the checkout process.
I appreciate that as a business you want to grab an email address as soon as possible, but personally I think that if you ask for it before the user is actually committed to the purchase, you risk putting them off entirely.
Hi, it’s a different project but uses the same backend I developed, just with different integrations
The landingpage has the location listed, but it’s not yet public.
But yea you’re right, I should list the locations for canvay
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
Adding an ssh key doesn't seem to verify it. It'd be nice to not allow adding invalid keys, in case someone screws up the cut-n-paste and truncs it or something.
A really big entry in the ssh key description or public key leads to /keys with a 500 server error.
Node/options to change the hostname doesn't. It seems to wipe and rebuild the node instead!
At 7/16/2022 14:43:42 UTC 2a01:4f8:a0:4078::c00:1 keeps dropping my public key, like maybe it's being rebuilt every few minutes (just long enough for me to install some packages). Once was me, the others were not. Maybe you're shutting down the beta and that's what I'm seeing?
Can't seem to select multiple ssh keys when deploying a new node. Is that by design?
singapore & tokyo will make great locations for this service. India and nearby countries have a higher than average ipv6 adoption rate according to apnic stats https://stats.labs.apnic.net/ipv6/
Adding an ssh key doesn't seem to verify it. It'd be nice to not allow adding invalid keys, in case someone screws up the cut-n-paste and truncs it or something.
A really big entry in the ssh key description or public key leads to /keys with a 500 server error.
Node/options to change the hostname doesn't. It seems to wipe and rebuild the node instead!
At 7/16/2022 14:43:42 UTC 2a01:4f8:a0:4078::c00:1 keeps dropping my public key, like maybe it's being rebuilt every few minutes (just long enough for me to install some packages). Once was me, the others were not. Maybe you're shutting down the beta and that's what I'm seeing?
Can't seem to select multiple ssh keys when deploying a new node. Is that by design?
I've added SSH key validation to my list, thank you!
Updating the hostname should be fixed, sorry, I was referencing the wrong form.
I also checked the key problem you mentioned, and located the issue, I made some networking changes yesterday and it looks like there's an edge case that can result in two VMs getting the same prefix.
I'll report back once it's fixed
The beta will stay active for a few more days but I'll switch the backend from the dev branch to production next week
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
@JBB You should see a new prefix if you rebuild your node, In case you've added some domains you probably have to recreate them due to the IP change.
Sorry for the issue, two VMs got deployed a couple of seconds apart, I'm currently working on implementing another check to make sure that won't happen again
Edit: IP assignment should be working again
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
Sorry for the radio silence recently, I've been buried at work.
I did play around with a hacked up version of zerofree, looking to see if there was any interesting leftover data in the unallocated blocks on the disk. Other than some apt cruft, it was clean. (And yeah, I did fill my disk with X, rebuild, and see if any data survived the rebuild. It did not.)
Planned on banging around on the ticket submission stuff a bit, but never got around to it. Sorry.
Couple of comments on nice to have stuff:
VNC: having been burnt in the past with unauth access to my VNC console, I strongly prefer the ssh tunnel setup that someone (Oracle? AWS?) uses. No HTML5 VNC console, but it just seems so much safer. (To be fair, I've never tried to ssh tunnel to other port numbers at Oracle or AWS, but surely they thought of that?)
It's more of a problem with IPv4, but I have noticed that with some providers I see a ton of (non-broadcast) traffic on my interface that doesn't involve my IP or MAC. Screws with my vnstat and munin traffic monitoring. So if you have a way to filter out traffic by guest (maybe each guest instance has an isolated vlan?), that's a good thing.
My favorite console access is the way tornadovps (nee prgmr.com) does it. ssh keys only, then a pick-a-number front end to select the serial console, rDNS, add/remove keys, reboot, shutdown, etc. Secure, fast, low bandwidth, and cut-n-paste works in the console.
Comments
Thank you for your extensive testing!
yea I’m aware of the pricing differences, most plans currently have the wrong price/ plan ID configured so stripe shows the pricing of a different plan.
I’ll create the final plans later and set the proper IDs.
The failed checkout attempts should get cleaned automatically after a few days, but ill add a button to delete them manually.
The goal is to assign a routed /56 to each node, but that might not be possible in each location.
Yes, the nodes have a NAT64/ DNS64 namserver configured by default to allow access to IPv4 only resources like GitHub
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
Yep totally agree, I'll add a info text or find a way to query caddy for the current ssl status.
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
I didn't try your beta, just clicked on the link in your signature, which I guess is the same project.
It doesn't seem possible to determine what regions you provide services in without going through the ordering process, and the first step of that requires you to register. Maybe it's just me, but as a user, I prefer to visit the order page a couple of times checking out the options etc, maybe coming back in a few days, before actually committing and I'd expect registration to be integrated in the last step of the checkout process.
I appreciate that as a business you want to grab an email address as soon as possible, but personally I think that if you ask for it before the user is actually committed to the purchase, you risk putting them off entirely.
Hi, it’s a different project but uses the same backend I developed, just with different integrations
The landingpage has the location listed, but it’s not yet public.
But yea you’re right, I should list the locations for canvay
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
Spoofer test: https://spoofer.caida.org/report.php?sessionkey=iskfh25w0djzfp (looks good!).
Adding an ssh key doesn't seem to verify it. It'd be nice to not allow adding invalid keys, in case someone screws up the cut-n-paste and truncs it or something.
A really big entry in the ssh key description or public key leads to /keys with a 500 server error.
Node/options to change the hostname doesn't. It seems to wipe and rebuild the node instead!
At 7/16/2022 14:43:42 UTC 2a01:4f8:a0:4078::c00:1 keeps dropping my public key, like maybe it's being rebuilt every few minutes (just long enough for me to install some packages). Once was me, the others were not. Maybe you're shutting down the beta and that's what I'm seeing?
Can't seem to select multiple ssh keys when deploying a new node. Is that by design?
Tokyo please
singapore & tokyo will make great locations for this service. India and nearby countries have a higher than average ipv6 adoption rate according to apnic stats https://stats.labs.apnic.net/ipv6/
I've added SSH key validation to my list, thank you!
Updating the hostname should be fixed, sorry, I was referencing the wrong form.
I also checked the key problem you mentioned, and located the issue, I made some networking changes yesterday and it looks like there's an edge case that can result in two VMs getting the same prefix.
I'll report back once it's fixed
The beta will stay active for a few more days but I'll switch the backend from the dev branch to production next week
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
@JBB You should see a new prefix if you rebuild your node, In case you've added some domains you probably have to recreate them due to the IP change.
Sorry for the issue, two VMs got deployed a couple of seconds apart, I'm currently working on implementing another check to make sure that won't happen again
Edit: IP assignment should be working again
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
The beta will end today, thank you everyone for testing!
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
Sorry for the radio silence recently, I've been buried at work.
I did play around with a hacked up version of zerofree, looking to see if there was any interesting leftover data in the unallocated blocks on the disk. Other than some apt cruft, it was clean. (And yeah, I did fill my disk with X, rebuild, and see if any data survived the rebuild. It did not.)
Planned on banging around on the ticket submission stuff a bit, but never got around to it. Sorry.
Couple of comments on nice to have stuff:
VNC: having been burnt in the past with unauth access to my VNC console, I strongly prefer the ssh tunnel setup that someone (Oracle? AWS?) uses. No HTML5 VNC console, but it just seems so much safer. (To be fair, I've never tried to ssh tunnel to other port numbers at Oracle or AWS, but surely they thought of that?)
It's more of a problem with IPv4, but I have noticed that with some providers I see a ton of (non-broadcast) traffic on my interface that doesn't involve my IP or MAC. Screws with my vnstat and munin traffic monitoring. So if you have a way to filter out traffic by guest (maybe each guest instance has an isolated vlan?), that's a good thing.
Good luck with the roll out!
Thank you for very much for your feedback!
VNC will get integrated in the next few weeks
AMD EPYC / NVMe / 10GBPs KVM in Frankfurt - https://v6node.com
Looking for an unbeatable AMD EPYC Baremetal Server in Frankfurt? Drop me a PM
My favorite console access is the way tornadovps (nee prgmr.com) does it. ssh keys only, then a pick-a-number front end to select the serial console, rDNS, add/remove keys, reboot, shutdown, etc. Secure, fast, low bandwidth, and cut-n-paste works in the console.