[SOLVED] Implement hot snapshots


#41

up. @PierreAntoine / @edouard can we get some answers?


#42

Hello,

We are still developping new offers and services!

For exemple, recently, we were the first ones to come out with ARM64 offers!

We also came out with the x86-64 offers, you can now Scale from 2.99€ to 34,99€ a month servers.

We are defenetly, only at the begining of Scaleway!

So, no, we did not stop development and there will be plenty of new surprises to come out.

In a short term we are concentrating in deploying more servers and volumes in the datacenter because this is the first issue we are having, also deploying new bandwith.

Plenty of little things that unfortunetly you do not see as much as you would see if it were to be new features, but we are of course, keeping the development of Scaleway

Best regards,
Pierre-Antoine


#44

Thanks, but why is there no changelog then, or a small news page? would be sufficient to just have a little page to see there is something happening and see what is planned / in progress.

Still there are too many open questions, since months / years:

Why are images not maintained?
Why is SIS not available since several years?
Why is bandwidth a reoccuring issue? been here less than a year, had issues with bandwith / loss too many times.
Why is there no proper IPv6 support?
Why do server keep failing to reboot / take hours or even days to reboot? (pain in the a** if you just want to add a new volume)
Why is there still a new random internal IP after an archieve (to add a volume), why can’t we have sticky internal IPs?

Why does it take ages to update your status pages? I had to annoy online.net a few times on twitter to get any information, since you do no update anything by yourself.

Edit: Right here in this thread, why is there no update / answer for almost 3 YEARS?! there are a lot of people in favor of this feature.

Why are abuse reports not handled properly or even ignored? -> Abuse reports ignored?


#45

Why are images not maintained?

  • We do work on giving you the latest versions of the OS, but indeed, we need to work on updating premaid images.

Why is SIS not available since several years?

  • SIS is still under testing, we do not want to release a not entirely functionnal or buggy service

Why is bandwidth a reoccuring issue? been here less than a year, had issues with bandwith / loss too many times.

  • Again, this is worked on, as you probably saw, we are regulary updating the network, but we are facing high demands.

Why is there no proper IPv6 support?

  • IPv6 is functionnal, we are listening to your feedbacks about it

Why do server keep failing to reboot / take hours or even days to reboot? (pain in the a** if you just want to add a new volume)

  • As said before, these are little things that we are updating regulary that you may not see on your side, it is not an easy thing to deploy and it is worked on very hard. Like the bandwith, the more we have customers, the more our plateform is overloaded. You probably also noticed, that archiving and reboots are faster than before already.

Why is there still a new random internal IP after an archieve (to add a volume), why can’t we have sticky internal IPs?

  • We had developed Scaleway with non fixed IPs for internal network at the begining. You can easly understand that to rewrite this, change the infrastucture is not a thing that can be done easily, but we heard your feedback about this also :wink:

Concerning abuses, non of them are ignored, they ALL need an answer from the customer. If there is no answer, the service is locked after 48 hours and the customer will need to explain us exactly what he has done to fix the issue before getting locked.

I hope I answered all your questions. The best would be to open a ticket if you have more questions as it will be faster to see them :wink:

Best regards,
Pierre-Antoine


#46

Thank you. Everybody should have an insight in what’s going on, so a public discussion is perfect.

I honestly do NOT believe that fixing SIS, since it was already public, or chaning the infrastructure takes almost 3 years. Same for hot snapshot implementation.

Random assigned ipv6 addresses without the ability to reserve / reuse them is NOT functional at all. It’s a mess.

Failing reboots and extremely long archive / shutdown times are not “little things” as you state. Those are very basic features, every other competitor manages to handle those kind of things with ease, just scaleway struggles a lot.

And I did not notice, beacause I fear rebooting my instances, due to the fact I would have to wait several hours / days to get it back up.

You managed to have sticky / reserved external IPs, why not internal. Shouldn’t be that big of a deal tbh, since you had several years to develop.

You still did not answer the questions regarding you information policy. Why is there no transparency regarding your ongoing progress and information policy?


#47

Why is there no proper IPv6 support?

  • IPv6 is functionnal, we are listening to your feedbacks about it

If IPv6 is functional, I suggest you make IPv4 work the same way it does and see if you keep your customers.

Why is there still a new random internal IP after an archieve (to add a volume), why can’t we have sticky internal IPs?

  • We had developed Scaleway with non fixed IPs for internal network at the begining. You can easly understand that to rewrite this, change the infrastucture is not a thing that can be done easily, but we heard your feedback about this also

You have over 17 million addresses in private space at your disposal. Do you have 17 million machines vying for this address space?

Both IPv6 and internal IPv4 seem like they’ve been designed with an assumption that you would be limited in space when in reality you aren’t possibly that limited.

Here’s a question for @PierreAntoine: How many public facing IPv4 addresses and IPv6 addresses are currently allocated for use by Scaleway but sitting there completely unused?


#48

Each Server probably got allocated its own internal IP. As you get a new one when switching, you also get a new IP.

My idea for this would be to allow users to shut down their server without uploading everything. In a nutshell, the server stays allocated to the user with its volumes. You can also just switch the computing instance, but the volumes stay local.
To prevent abuse, you can set automatically clear the allocation after 1 hour or something like this.

The quick scalability is important, and currently not 100%ly provided. Scaleway is very interesting for smaller customers, as the servers are cheap and easy to upgrade without having to setup everything again
@PierreAntoine


#49

up. give us a heads up on this, since scaleway is failing again… Same issues over and over again. No promised improvements “behind the curtains”. Seems like a lie yet again.

It just get’s worse. See AMS since days / weeks.


#50

No hot snapshots yet?


#51

nope, no hot snapshots yet.


#52

I’m super annoyed that this is not a thing yet.
My entire production system runs on scaleway and a reboot just to create a convenient backup is annoying as hell…


#53

+1 for virtual machines it is must have


#54

Hot snapshot are available in preview and GA is scheduled in the next coming days! https://blog.online.net/2018/05/03/introducing-scaleway-nextgen-nvme-cloud-servers-with-hot-snapshots/


#55

Eduard, to make it clear:
can I convert old-style VC1S to new START1-S?


#56

Yes, at the moment it’s only possible in Paris, It will be possible in Amsterdam in the next few days.
To upgrade your VC1S, archive your server and then power it on. You will start on a start1-s cloud server.

Note that the commercial type of your server will remain the same VC1S after this operation.


#57

Wow, just 3 years and 5 months, great job…

When will hot disk add remove be available and/including server shutdown without transfer? Another 3 years?


#58

Hi @gentlemon,

We plan to release our roadmap publicly to help you understand what we are working on and the features / products planned for next months.

Ed


#59

Finally! About time you do that.


#60

It took some time to get this feature, sure, but please don’t forget the incredible amount of work that has been done with in-house electronics to bring us these small servers at such incredible prices on a range of architectures. Let’s give the team some credit?


#63