Ingress_expiry not within expected range error

We are not 100% but, there might be a bug. So if anyone can give a hand test an hypothesis please follow-up there:

It does not fix the problem…

A pity :confused:. Can you share your more details, like source code (if open source) and link to your app on mainnet? Assuming my colleagues may be interested to have a look.

Not open source for now, sorry(
I am developing ledger scanner and it has to constantly call get_blocks or get_transactions. I noticed that every ~20 minutes this issue appears. My program automatically restarts broken process and requests work fine for ~20 minutes.
I send the source code here: Ingress Expiry Issue - Your Input Needed!

1 Like

Thanks. I have share your feedback.

1 Like

@yvonneanne @peterparker @jennifertran my website is about article when i try to publish the article then i face this error but when i change i pc time 1 mint up then its published and when i set it automatically then i face this error , issue is that mostly preople in world has time automatically set and i can’t ask all of them to set there time 1 mint up, kindly solve this error,
Specified ingress_expiry not within expected range: Minimum allowed expiry: 2024-09-25 16:20:14.229894131 UTC, Maximum allowed expiry: 2024-09-25 16:25:44.229894131 UTC, Provided expiry: 2024-09-25 16:20:00 UTC

Thanks for the feedback. The SDK team has reverted a feature and released a new version of agent-js (v2.1.2). Could you give it a try and let them know if it helps?

i am already using older version of becasue my project is 1 year older
@dfinity/agent”: “^0.18.1”,
@dfinity/auth-client”: “^0.18.1”,
@dfinity/candid”: “^0.18.1”,
@dfinity/identity”: “^0.18.1”,
@dfinity/ledger-icp”: “^2.0.0”,
@dfinity/nns-proto”: “^1.0.0”,
@dfinity/principal”: “^0.18.1”,
@dfinity/utils”: “^1.1.0”,

Oh I see. I’ll forward your feedback to the team.

It doesn’t seem like the issue is related to agent-js; this looks more like a backend problem. In our case, inter-canister calls are also failing, including one-way calls that are ending up unsuccessful.

So I redeployed OpenFPL to new canisters, surely they are on a new subnet than the old one I deployed to a year ago, I get the same error from a new canister:

1 Like

Hi James. We think this issue is a compound of subnet load issues, along with the JavaScript agent’s attempt to reconcile differences between replica time and system time.

Since it doesn’t appear that subnet load was the factor in your case, I suspect it may be the agent version. Can you confirm if you are using version 2.1.1? If so please update to 2.1.2 or downgrade to 2.1.0

To be transparent, we are working on a protocol-level solution that will allow agents to synchronize clocks with the subnet they are calling into in a trustable manner. These ingress expiry errors will go away once we have that up and running. We should have something in beta in a couple weeks.

Super sorry for the trouble and inconvenience. We probably could have solved it with an unsafe hack, but that’s not what we’re about. I hope you and others can accept that we are putting more time and effort into this so that we ship something safe and reliable.

Can I gift you a Cycles coupon (or something else) for your trouble?

2 Likes

Gift me some cycles lol. I dunno how to get across the money involved in this for me.

So I take the ICP I raised in the SNS, pay developers and associated other staff, we don’t have an app to promote.

Everything has a lead time, so repeatedly, the string of delays I’ve experienced have cost me tens of thousands of pounds.

It’s worse than that, because I truly decentralised my app, I have to go to my DAO token holders and ask for more funding, on the back of not delivering. Thanks for this, I’m sure they’ll be keen.

I have a backlog of campaigns, content and other stuff that just sits here, all just wasting away. Like this week I wanted to bring out the women’s game, OpenWSL, a new women’s football neuron, I have people associated with the women’s England team ready to be involved, I have sponsorship deals ready to sign.

The pressure of having people leave their web2 jobs to join the team, I had no idea this was some test project not ready for production.

So no, I don’t want cycles, I want you to recognise you are potentially destroying my life, and for what? AI? 64 bit? Bob.fun blocking up the subnets, if only you’d supported OpenFPL the way you supported that ICP drain.

1 Like

I get it, and I can only imagine what being under that kind of pressure is like.

Every platform, every cloud provider has imperfections. ICP is a relatively small but growing project that is getting better every day. What cloud provider is going to have their engineers and executive leadership reach out to you directly? What cloud provider is going to prioritize your problems and tailor their roadmap to make sure you’re satisfied? Our commitment is to you and all of the developers building on ICP. So while I acknowledge you are experiencing some pain, I assure you that your forbearance will pay dividends in the end.

And to get back to the issue — can you tell me your agent version so we can validate our hypothesis and run more tests internally?

I was 2.1.1. Upgraded to agent 2.1.2

Just making note of the date and time as I have data but i’ve had data before, hopefully this agent package update has fixed things for the time being.

and 20 mins later I can no longer run that:


And now running it is back to full ingress_expiry error

Under half hour.

I think its not related to version because my project is one year old and using older version of definity , it am facing this issue from last 3 weeks. I can do nothing. As l already told that my website is about articles, i am unable to create article if i put my time 1 mint ago then its work.

I’ve also been stuck on this issue for quite a few months now. I’ll take a cycles coupon and cheeky temporary work around if thats available. Thanks

Advice I’ve received from community member is to:

up the compute_allocation on the canister_settings to 1, will burn more cycles but probably fixes the issue.

Will try when I figure out how

1 Like