search

BuildOnViction / tomox-sdk

SDK to build a Decentralized Exchange on TomoX protocol
MIT License
36 stars 18 forks source link

order disappear, can't cancel and funds locked in orders #149

Closed Luc-Michault closed 4 years ago

Luc-Michault commented 4 years ago

Hello, a lot of Open Orders are disappearing after few days, so we can't cancel them anymore.

For example : https://scan.tomochain.com/orders/0xb225ea0e18c70b354c18770eff4d9a7ccd83974919f4bf3358c6303d2d0a8ac3 an order i put on 2 months ago :

order

As you can see, order is appearing on orderbook while it's absent from Open orders, so i can't cancel it.. I don't see them locked in VisioDEX but if i log in TomoDEX, those tomo are locked in orders

stuck in orders

So my wallet can't interact on Tomo paires without at least 379.73 + 1 Tomo

Someone else complains about this situation with a stuck open order which disappeared : https://scan.tomochain.com/orders/0x1132b29b1788275d5524c012e6b05637ed36ca2c25b9245f62d197a9e8b1d36e

It was only from 2 days ago

thanhson1085 commented 4 years ago

Ok, we will investigate it.

Luc-Michault commented 4 years ago

Any way to cancel those order manually while you investigate ?

thanhson1085 commented 4 years ago

you can check your mongo database, make sure that the order is stored in db.

thanhson1085 commented 4 years ago

Seems NOT, https://trade.visiodex.com/api/orders?address=0x42db3fd6a0c68389a1935d1bff6479172bcb7426&orderStatus=OPEN

But It shows here: https://dex.tomochain.com/api/orders?address=0x42db3fd6a0c68389a1935d1bff6479172bcb7426&orderStatus=OPEN&relayerAddress=0x83c7399d01e3e1c20e5ffd7957c8666d73c1ad9b

Your DB missed the order, you need to re-run the fullnode mongodb

Luc-Michault commented 4 years ago

Ok, how i proceed? i just delete the docker volume and reinstall it? Why those orders disappeared from DB ?

thanhson1085 commented 4 years ago

Run from scratch in the new server, then point the domain to the new server

Luc-Michault commented 4 years ago

from scratch without snapshot ?

thanhson1085 commented 4 years ago

Run quick-start script, then remove tomox datadir, restart fullnode

Luc-Michault commented 4 years ago

Ok that's what i did, 4millions block to go now.

We need an alternative to fix this kind of issue, what would happened in few months/years when blockchain will be way more heavy? it will be impossible to stop node for days to catch 10-20-100 millions block.

I thought i was safe with my chaindata snapshots but obviously, i'm not.

Now i understand why older order disappeared as we migrated to new server a month a go, but it doesn't explain why this order https://scan.tomochain.com/orders/0x1132b29b1788275d5524c012e6b05637ed36ca2c25b9245f62d197a9e8b1d36e disappeared / was missed by DB.

Edit :

Maybe you could make a tool to resynchronize mongodb, as reading again all tomox data to actualize stored data and verify integrity of orders, i'm sure it would be way faster then resynchronize millions of block from node. Is it possible ?

I'll do from scratch today, but we need better solution for a possible next time.

Luc-Michault commented 4 years ago

Hello, i need another solution as node already had 2 rollback to 20m block, it's already sync for 4 days without any results so i think rollback issue is still not fixed on nodes.

I'm stuck, can't use snapshot to avoid empty database, but sync is not working, what i'm supposed to do..?

thanhson1085 commented 4 years ago

If you want to have full orders, trades db, the fullnode have to re-sync the data from the first tomox block.

If the node can not re-sync, let us know more logs

Luc-Michault commented 4 years ago

Hello, i can't provide more logs since logrotate are deleting them, how i'm supposed to know that node rollback, specially when it happened when i'm asleep... That's why i already told you to clear those error logs and keep only true errors, it's not called error log to catch every "debug" informations, which means every actions... those error logs are unusable in such state...

As i said, i need another solution then "reinstall everything" or "re-sync the data from first tomox block", will i need to do that every time there is a tiny issue? Do i need to shutdown business for a week every time ?

Can you please dump a backup of mongodb ?

thanhson1085 commented 4 years ago

You do not need to shutdown business to fix the minor issue.

The old server is OK with the minor bug. You run a new server, setup the re-sync. When the new server is ready, you switch the domain to the new one with zero-downtime.

I do not have mongodb backup for now.

I suggest you a solution to save the time. You just need to run the frontend in your server and use our backend. So you do not need to run fullnode, mongodb, sdk, you just need a small server to run the frontend.

Luc-Michault commented 4 years ago

Ok i'll try that way until i can fix that node issue, so people who are stuck can finally remove their order.

How do i use your backend ?

thanhson1085 commented 4 years ago

2 steps:

Let me know the domain when you finish the step one. I need to know the domain do config our backend.

Luc-Michault commented 4 years ago

Ok the domain is api.visiodex.com, it's done. I'll update FE.

thanhson1085 commented 4 years ago

OK, you go relayer.tomochain.com, unlock the owner wallet. and update the dex url info to https://api.visiodex.com, it should be work automatically.

thanhson1085 commented 4 years ago

If socket does NOT work, try to update config:

REACT_APP_ENGINE_WS_URL='wss://api.visiodex.com/socket'
Luc-Michault commented 4 years ago

My associate is on vacation so i'll have to wait him to change relayer information.

Still trying to sync the node, it's really really slow, log is full of :

9|tomox-fullnode | DEBUG[08-28|11:56:35] Discarded propagated block, too far away peer=f9ff0f88c8f2b39e number=25026866 hash=859477…0031c8 distance=1902449

9|tomox-fullnode | DEBUG[08-28|11:56:35] Discarded propagated block, too far away peer=0f0d344b874fbbf0 number=25026867 hash=f736ec…975f7a distance=1902450

9|tomox-fullnode | DEBUG[08-28|11:56:35] Discarded propagated block, too far away peer=917f8dc30d9247a8 number=25026867 hash=f736ec…975f7a distance=1902450

9|tomox-fullnode | DEBUG[08-28|11:56:35] Discarded propagated block, too far away peer=e69e47bb05914418 number=25026867 hash=f736ec…975f7a distance=1902450

9|tomox-fullnode | DEBUG[08-28|11:56:35] Discarded propagated block, too far away peer=12958408819f4df3 number=25026867 hash=f736ec…975f7a distance=1902450

9|tomox-fullnode | DEBUG[08-28|11:56:35] Discarded propagated block, too far away peer=f51cd6ea9683a608 number=25026867 hash=f736ec…975f7a

Luc-Michault commented 4 years ago

Relayer information are modified, but impossible to load a pair on spot trading when i connect with metamask or with tomowallet on dapp, it displays "Cannot read property 'baseTokenDecimals' of undefined".

thanhson1085 commented 4 years ago

Did you update the .env and rebuild FE?

REACT_APP_ENGINE_HTTP_URL='https://api.visiodex.com/api'
REACT_APP_ENGINE_WS_URL='wss://api.visiodex.com/socket'
Luc-Michault commented 4 years ago

Did you update the .env and rebuild FE?

REACT_APP_ENGINE_HTTP_URL='https://api.visiodex.com/api'
REACT_APP_ENGINE_WS_URL='wss://api.visiodex.com/socket'

Yes

thanhson1085 commented 4 years ago

Relayer information are modified, but impossible to load a pair on spot trading when i connect with metamask or with tomowallet on dapp, it displays "Cannot read property 'baseTokenDecimals' of undefined".

It works in my side. I think you should clear the cache by change the version in .env file:

REACT_APP_DEX_VERSION=xxx
Luc-Michault commented 4 years ago

Did you try to login? it's not working after being logged on.

I changed the version in .env file, but it's not working.

thanhson1085 commented 4 years ago

Yep, logged in Ok with metamask. You should create another issue with the details, we can investigate more. thank you.