Open Wrive opened 2 years ago
coco98
commented
2 years ago Hey folks. Thanks for reporting this behaviour.
Could you share more details that would help us investigate / reproduce the behaviour?
Some things that would help: The query itself. The difference between 1.3 and 2.0 sql generation of any. The amount of data being selected. And where you’re currently running the hasura instances.
Wrive
commented
2 years ago Hey there, we are running Hasura in Azure. v2 in AKS and v1.3 on a dedicated VM. The query is complex and when the query hits a large number of records this is where we run into problems. When the payloads are small-medium, v2 is up to the task and no problems.
If you would like to email me directly, i can share more info with you offline - paul.doherty@wrive.io
sassela
commented
2 years ago Thanks for raising this @Wrive, I've emailed you to request more info
Jeancr123
commented
2 years ago Are there any updates on this? The timing out of the API operations makes things very complicated
Wrive
commented
2 years ago Hi Team
This is still a problem for us too. We have had to manage it through chunking data size from our API calls. We would like to use Hasura in other projects, but for now, don’t see it as a viable, enterprise solution for us, until we are sure of its stability as there is no error handling for long queries.
I would be happy to spend some time working through this with the team to identify root cause and a solution.
Regards,
Paul Doherty Enterprise Architect +61 457 929 193
[Icon Description automatically generated] Level 1, St. Kilda Road Towers, Email: @.**@.> 1 Queens Road, Melbourne, Phone: +61-3-9016-9393<phone:+61390169393> Victoria, Australia, 3001 Website: wrive.iohttps://aus01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwrive.io%2F&data=04%7C01%7Czeon.joseph%40deepbluecompany.com.au%7C9898451bdab34106e0db08d9daf61872%7C20662f19dcf543db861d23acdd8eeb44%7C0%7C0%7C637781574531546248%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=MLfSTMghE%2B4%2Fo50NLkz%2BoNRUalAnMVSIQwG5brLoqtg%3D&reserved=0
GET OUR FREE DIGITAL STRATEGY & INNOVATION ROADMAP TO START YOUR DIGITAL TRANSFORMATION TODAY: wrive.io/digitalstrategyhttps://aus01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fwrive.io%2Fdigitalstrategy&data=04%7C01%7Czeon.joseph%40deepbluecompany.com.au%7C9898451bdab34106e0db08d9daf61872%7C20662f19dcf543db861d23acdd8eeb44%7C0%7C0%7C637781574531546248%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000&sdata=lONoGr6OjpcxJigmJfumr22wh3pkGjn%2FJu7VdJoqjgg%3D&reserved=0
From: Jeancr123 @.> Sent: Tuesday, 10 May 2022 8:49 AM To: hasura/graphql-engine @.> Cc: Paul Doherty @.>; Mention @.> Subject: Re: [hasura/graphql-engine] v2.0.3 long queries (#7464)
Are there any updates on this? The timing out of the API operations makes things very complicated
— Reply to this email directly, view it on GitHubhttps://github.com/hasura/graphql-engine/issues/7464#issuecomment-1121658476, or unsubscribehttps://github.com/notifications/unsubscribe-auth/AVH34ODWRJPFV5P5QSUFCV3VJGI6TANCNFSM5CZHULPQ. You are receiving this because you were mentioned.Message ID: @.**@.>>
Hey guys, we are running versoin 1.3 and version 2 pointing at the same PostGreSQl DB. When we run a query that happens to return a lot of data, Hasura 2.0.3 just bails on it. No error, no timeout, just nothing in the GraphQL payload window.
However, when we run exactly the same query in version 1.3, pointing at the same DB, it does return the payload. When we use the Analyse feature to get the SQL query and we run this on the DB directly, there is only about 1-2 seconds difference in the return time for the payload in SQL compare to Hasura 1.3.
however v2 just bails on the query and does nothing - we have found this to be the case after multiple tests. For queries that exceed 1 min (or thereabouts) version 2 does nothing.
We have 1.3 in dev and 2 in prod and would prefer not to have to rebuild to 1.3. We have also indexed the tables in PGQSL following the how to guide on improving query performance (but from the above, that was really irrelevant and made no difference)
We are running 2.0.3 in an Azure container pointing at an Azure database for Postgres SQL. 1.3 is running in docker desktop on a VM
Was difficult to find any support from Hasura for self managed instances (other than GitHub). A support chat feature on your website wouldn't go astray.