Guidance needed on solution Expose - no gateway available (following Hugo website tutorial)

Hi everyone,

I’m following the Hugo tutorial (https://manual.threefold.io/#/tutorial_hugo_website_advanced) on testnet and everything went well so far, but I’m stuck on the solution expose part…

I have set up my farming node in my VM, created a capacity pool and the network to access it, the container works well on local net accessible by wireguard. When I try to add the solution expose, after selecting the exposed ports, I’ve got an error message : “no gateways available in your pools.”

What gateway is it talking about ? I didn’t see any documentation about this !

Any guidance welcome ! :slight_smile: Thank you !

1 Like

Hi there, webgateway is one of the core elements of the grid it’s like a portal that let’s you get in and talk to the workloads deployed on the private network. Its code exists here. And to expose something to the outside world, you definitely need to have a pool with a gateway -you run yourself or someone else running it- added to it. on testnet there is a gateway on freefarm you can use.

1 Like

Hi Ahmed, thanks for your reply !

As I’m trying to deploy on my capacity pool, I apparently cannot use the gateway from freefarm, it’s not selectable, the error message is pretty clear that my capacity pool does not have gateway (or maybe I’m missing something)

I saw your github link and would like to run my own gateway, but I don’t see how I can deploy it on my testnet farm - any tutorial available for this ?

Another question : I saw on this thread (How do save Data (like SQL) on HRU Capacity?) that it is possible to create a local volume based on SSD or HDD, how can I do that ? Are these volumes also “dispersed store” as described here (https://manual.threefold.io/#/architecture_storage?id=threefold-space-algorithm-dispersed-storage) to be sure not to loose data ?

Thanks for your guidance, I’m eager to learn more about threefold technology :slight_smile:

Hi @naturecrypto the idea is you can utilize gateways of any farm, so if you provide the compute / storage nodes on your farm, you can still use gateways from other farms. Just create a pool and add to it the gateway node on freefarm :slight_smile:

For deploying your own gateway I’ll ask @delandtj to provide you with the right links for deployment

I double checked with @azmy for the most accurate answer. dispersed storage is only available via our s3 storage solution.

Hi @ahmedthabet thanks for your replies !

I added a small capacity pool (1 CU & SU) on freefarm and the solution expose workflow showed me the 3 gateways of the freefarm solution.

I tried all three of them and had the same error :

“Failed to reserve reverse proxy workload 504471. Please use the refresh button on the upper right to restart the chatflow”

The alerts tab showed more details : “Failed to decrypt secret: decryption error. For more details: https://explorer.testnet.grid.tf/api/v1/reservations/workloads/504471

Do you have any idea what went wrong ?

Also I’d be interested to know how to deploy local volumes, and if it is possible to make them failure redundant (for example configuring the storage layer as RAID 5 ?)

The final idea after my tests is to host my nextcloud server and mail server on threefold

Thanks a lot for your help much appreciated !

@naturecrypto Hi :wave: this error means you’re using an incompatible version, I suggest you update your sdk, the encyption scheme got hardened, and using the old release against testnet won’t be compatible

for the redundancy question, I see it in another thread and will leave the fruitful discussion to continue there

Hi @ahmedthabet,

update was no easy with pip : the actual js-sdk (11.0b13) needs criptography 3.0, but anoter dependancy (pyopenssl 20.0 or certbot-nginx) needs cryptography>=3.2.

So it’s necesseray to install first js-sdk, then certbot-nginx which upgrades criptography to 3.2, and then the sdk can start properly.

The last version could not list the capacity pools, got this error message : module ‘jumpscale.sals.zos.pools’ has no attribute list

In the console there is also "jumpscale.sals.zos.workloads’ has not attribute list

Another attempt with the poetry method in another venv gives the same result…

The problem does not appear with the 11.0b8 version

Any help appreciated :slight_smile:

Hi,

I tested again a few days ago and I had the same result… Did I do something wrong ? The upgrade process does not use pip or poetry ?

If I could have some guidance today I could work on it this week end, would be great !! :smiley:

Hello @naturecrypto

Sorry for the inconvenience. I created an issue to ensure we don’t face that again https://github.com/threefoldtech/js-sdk/issues/1786

for your issue, would you try cloning js-sdk , poetry install and that’s it (if you’re installing system wide it might be problematic -you need to uninstall first-

And for your error jumpscale.sals.zos.workloads’ has not attribute list can you share which document are you following? there was a recent update to zos sal and now all APIs require get call first j.sals.zos.get()

Hi @ahmedthabet,

I installed two fresh VM : Debian 10 and Ubuntu 20.04.1 LTS. Same result than on my manjaro.

For the two of them, installing js-sdk by pip or cloning js-sdk repo and poetry install does not complete. The steps are easily reproducible :

  1. Install fresh Linux distrib of your choice with all requirements
  2. Create a venv and try to install with pip js-sdk. Failure.
  3. Deactivate the venv, clone js-sdk repo and poetry install inside it. Failure.

All installations procedures fail with the last js-sdk version

For the error 'jumpscale.sals.zos.workloads’, if I remember correctly, it appears on the console when I go to the 3bot admin menu capacity pools, on the badly installed last version of js-sdk. A clean install of js-sdk with all the dependencies correctly satisfied would clearly not throw this kind of error, as I could have seen on a hosted 3bot deployed by 3bot Deployer. I hope you will resolve this dependency issue, for the moment I will use the 3bot deployer :slight_smile:

JS-SDK 11.0b14 installed succesfully with latest pip, no dependencies problems !

BUT when you try to get a Testnet funded wallet, http://gettft.testnet.grid.tf/tftfaucet/api/transfer times out…

@ahmedthabet Could you check your API to get some testnet TFT please ? :slight_smile:

Payment has been changed for testnet/devnet since this release, testnet TFTs are not used anymore, all payments are done with real TFTs (but for 90% discount for testnet and 99% discount for devnet).

A new release that reflects these changes will be out and install-able via pip soon, until then, you can use install js-sdk from development branch (manually using poetry).

So there is actually no way to test your tech without paying ? Really ??

Yes @naturecrypto, at this point in our journey we believe that with the commercial rates for capacity as they are and a 90 to 99% discount it’s a good deal to have capacity available to develop on testnet and devnet. Tokens are available at some exchanges for a very low prices while in the system we use a higher token value hence the number of tokens required is minimal.

If you have a project that you think deserves funding, please write a project proposal and share it with us. We will then see if we can sponsor the project. Also we are working on a general bounty program and it is coming soon. Also we are working on a program for have recognized so called experts that can help the community and create a token income that way.

My intent is to farm, self host my IT needs and rent available capacity to other people, and probably create light hosting services on the TF grid, because I love the philosophy of this project and I want to support you and I think it could lead to a better future for us all

For this I need to test kubernetes deployment and ceph cluster creation on very small scale (or other solution of shared storage not object oriented) for high availability purposes. Testnet was really useful to work on for testing your tech before investing much time money and effort on a production release. It would allow me to gain some trust on your tech. However, when I began working on testnet a few weeks ago, I encountered some problems :

  • 3sdk was not installing properly with pip or poetry
  • Last version of zos does not provide IPV6 to a VM hosts anymore (I used my own VMs to provide capacity for my tests)
  • Kubernetes flist deployment was not working
  • No idea how the monitoring prometheus stack is working so I need to put my hands on it
  • No procedure for gateway deployment
  • Imprecise documentation at best and no technical deepdive available

With this kind of experience, the fact to have to pay to test my ideas on testnet is another blow to my motivation to invest time and money on the TF grid, even if it’s not that expensive, the fact to pay for temporary tests on testnet which can be rebooted without notice is not very comforting.

I’d appreciate a little TFT help to make possible the creation of my PoC and gain necessary trust on how your tech works !

2 Likes

Hi @naturecrypto,

I understand your frustrations and I really appreciate your perseverance. It is a massive project, understaffed and underfunded as we have a lot of people that go the extra miles to get this launched, stable and working. So we 100% appreciate any input, feedback and help we can get.

The step to go to a single currency is done for simplicity, fewer things operational means fewer things to create, operate and maintain. We already have a large number of components in this decentralised grid and currently, we are focussing on bringing this grid with a number of developers and end-user experiences to a larger audience.

Testnet has the latest version of the TF Grid, eVDC, 3bot and some more end-user experiences. Send me your wallet address and I will send you some.

Keep moving!

1 Like

I agree with this, still I think a public comm to say you’re switching to mainnet TFT for testnet and people could apply to a small TFT donation for test purpose would bring you some nice feedback !

Will do, thank you ! :slight_smile:

1 Like

Hi @naturecrypto
thank you for your feedback, we immediately went over and re-tested all the technical points you mentioned and we want to assure you that everything is working properly now.

you can find all docs on how to use them here:

please don’t hesitate to contact us if you experienced any further issue, and maybe you can provide more logs and error screen shots so we can help identify the issue.

Thank you for using our TF-grid; and please keep moving!

1 Like

I will ! Thanks for your answer.

I replied on the github issue

I can see your reply on github, thnx a lot.
I can also see your issue on zos had been resolved.

here will be nice if you want to post suggestions or questions about the docs.
or you can directly contact support via the TF connect app and they can help you with all technical issues.