this chart on the blog post gives me the impression that read replicas will pretty much always be un

this chart on the blog post gives me the impression that read replicas will pretty much always be under a second behind the primary.
is this impression accurate?

ajgeiss0702OP•4/10/25, 6:36 PM

also, when the sessions api was announced last year, IIRC it was stated that if you used the old (non-session) D1 api, the behaviour would be the same as it is without replication (always go to the primary) but now this graphic suggests otherwise. did this change?

Aajgeiss0702 also, when the sessions api was announced last year, IIRC it was stated that if ...

Chaika•4/10/25, 6:38 PM

the bottom left of that graphic "With using Sessions api"

ajgeiss0702OP•4/10/25, 6:38 PM

oh wrong graphic

ajgeiss0702OP•4/10/25, 6:38 PM

meant this one

ajgeiss0702OP•4/10/25, 6:40 PM

oh unless this is saying not using the session api (on a replica-enabled db) would be equivalent to using the sessions api with first-primaryfirst-primary?

Aajgeiss0702 oh unless this is saying not using the session api (on a replica-enabled db) wou...

Chaika•4/10/25, 6:43 PM

yea, I don't think the blog does a great job of explaining it, but you can see in the code itself, if just using the methods on the binding itself, it only goes primary
https://github.com/cloudflare/workerd/blob/04deab5ea50818faead9b18ce780d9a55ffccc9a/src/cloudflare/internal/d1-api.ts#L94

ajgeiss0702OP•4/10/25, 6:43 PM

ah ok, thank you for clearing that up for me!

ajgeiss0702OP•4/10/25, 6:45 PM

super excited to start using replicas!

Mozzy•4/10/25, 7:13 PM

Still getting D1_ERROR: Network connection lost.D1_ERROR: Network connection lost.

MMozzy Still getting `D1_ERROR: Network connection lost.`

ajgeiss0702OP•4/10/25, 7:36 PM

intermittently or every request?

ajgeiss0702OP•4/10/25, 7:37 PM

I get that too, but last I saw they were recommending people retry the request when reciving that error

Mozzy•4/10/25, 7:39 PM

occasionally, but, started a few hours ago. Could be read replication perhaps, but the error isn't helpful

ajgeiss0702OP•4/10/25, 7:40 PM

actually I just checked my logs, and ive seen more of those recently than usual too

ajgeiss0702OP•4/10/25, 7:40 PM

first one for me was April 10, 2025

Aajgeiss0702 first one for me was <t:1744311000:t>

Milan•4/10/25, 7:42 PM

timezone?

ajgeiss0702OP•4/10/25, 7:42 PM

i used discord's timestamp, which converts to your timezone

Aajgeiss0702 i used discord's timestamp, which converts to your timezone

Milan•4/10/25, 7:43 PM

Ah, that's why it was highlighted

ajgeiss0702OP•4/10/25, 7:43 PM

18:50 UTC

Milan•4/10/25, 7:44 PM

Any particular region?

ajgeiss0702OP•4/10/25, 7:44 PM

that might have been around when I enabled replication for that db, but i didnt record exactly when I enabled it so im not sure.
but the workers arent using the session api at all yet

MMilan Any particular region?

ajgeiss0702OP•4/10/25, 7:45 PM

the database is located in WNAM, let me check the colo of the requests with errors

ajgeiss0702OP•4/10/25, 7:48 PM

there doesnt appear to be a common colo. some in europe, some in asia, and a few more errors from YYZ and Chicago (which are the colos that my site gets the most requests from anyway)

ajgeiss0702OP•4/10/25, 7:49 PM

db id is

221100ba-5ab9-4cda-8cfa-69c17ea546ab

221100ba-5ab9-4cda-8cfa-69c17ea546ab

if it helps (account id is

f55b85c8a963663b11036975203c63c0

f55b85c8a963663b11036975203c63c0

)

Mozzy•4/10/25, 7:51 PM

Mine: Times are UTC. Errors are 99% "Network Connection Lost"

gruntlord6•4/10/25, 7:51 PM

I suddenly started getting this for a project that I am trying to redeploy as well:
binding DB of type d1 failed to generate. Please try again later or contact support. [code: 10021]

ajgeiss0702OP•4/10/25, 7:52 PM

at least for me, it doesnt appear to be every request that errors, just a handful. way more than before today though

gruntlord6•4/10/25, 7:52 PM

yea I was using D1 fine until just recently. I have seen this error before but normally clicking redeploy fixes it

Aajgeiss0702 db id is `221100ba-5ab9-4cda-8cfa-69c17ea546ab` if it helps (account id is `f55b...

Milan•4/10/25, 7:53 PM

I've let the team know

MMilan I've let the team know

ajgeiss0702OP•4/10/25, 7:53 PM

thank you

Mozzy•4/10/25, 7:53 PM

I did have one count of:

D1_ERROR: internal error; reference = s3pbr2bu1qcneucpdoi9h07c

D1_ERROR: internal error; reference = s3pbr2bu1qcneucpdoi9h07c

assuming the reference is not secret

Mozzy•4/10/25, 7:56 PM

Now also getting HTTP 500 from

https://dash.cloudflare.com/api/v4/accounts/account_id/d1/database?page=1&per_page=25

https://dash.cloudflare.com/api/v4/accounts/account_id/d1/database?page=1&per_page=25

datacenterdude•4/10/25, 7:57 PM

D1 go boom boom

MMilan I've let the team know

Zack•4/10/25, 7:58 PM

I'm also getting 500's on the D1 dashboard right now, checking for errors in production

Zack•4/10/25, 7:59 PM

I don't see any actual errors in production (worker with bindings), however the dashboard currently appears to be broken

datacenterdude•4/10/25, 7:59 PM

binding BOOKFRAME_DB of type d1 failed to generate. Please try again later or contact support. [code: 10021]

binding BOOKFRAME_DB of type d1 failed to generate. Please try again later or contact support. [code: 10021]

on a

wrangler deploy

wrangler deploy

MissS•4/10/25, 8:00 PM

Is it possible to create new databases dynamically yet? Or are we still stuck with being unable to on-the-fly create and interact with them?

Zack•4/10/25, 8:00 PM

{"errors":[{"code":7500,"message":"internal error; reference = e_28XXLS_1d6beafaa18e4d7985c51efb410cf88e"}],"success":false,"messages":[],"result":null}

{"errors":[{"code":7500,"message":"internal error; reference = e_28XXLS_1d6beafaa18e4d7985c51efb410cf88e"}],"success":false,"messages":[],"result":null}

Paul Shriner - Titleless•4/10/25, 8:05 PM

Hoping it is just a restart issue.

Tim•4/10/25, 8:07 PM

I'm also having issues with D1.

Dashboard

When trying to load my list of databases, the

GET https://dash.cloudflare.com/api/v4/accounts/X/d1/database?page=1&per_page=25

GET https://dash.cloudflare.com/api/v4/accounts/X/d1/database?page=1&per_page=25

returns a 500 like,

{
    "errors": [
        {
            "code": 7500,
            "message": "internal error; reference = e_28XXLS_d2cd6b5618884603b87fab149025c82e"
        }
    ],
    "success": false,
    "messages": [],
    "result": null
}

{
    "errors": [
        {
            "code": 7500,
            "message": "internal error; reference = e_28XXLS_d2cd6b5618884603b87fab149025c82e"
        }
    ],
    "success": false,
    "messages": [],
    "result": null
}

---

Deployment

My deployment of a "Page w/ Worker", which has a D1 binding is failing with the following,

Error: Failed to publish your Function. Got error: binding DB of type d1 failed to generate. Please try again later or contact support.

Error: Failed to publish your Function. Got error: binding DB of type d1 failed to generate. Please try again later or contact support.

nejc•4/10/25, 8:09 PM

Same problem here

nejc•4/10/25, 8:09 PM

{"errors":[{"code":7500,"message":"internal error; reference = e_28XXLS_75dd26695d58424fb92679a4a8ca2038"}],"success":false,"messages":[],"result":null}

nejc•4/10/25, 8:09 PM

I am getting this on all databases that I tried (and we have thousends of them)

Isaac McFadyen•4/10/25, 8:11 PM

D1 binding issue is being looked at by engineering internally, thanks for the reports.

Paul Shriner - Titleless•4/10/25, 8:12 PM

AND WERE BACK

MMissS Is it possible to create new databases dynamically yet? Or are we still stuck wi...

ajgeiss0702OP•4/10/25, 8:15 PM

dynamic bindings are not released yet. Until they are, Durable Object's SQL storage is something you could look into, since you can create an unlimited number of objects, and each has seperate storage.
Note: https://developers.cloudflare.com/durable-objects/best-practices/access-durable-objects-storage/#sql-in-durable-objects-vs-d1

Aajgeiss0702 this chart on the blog post gives me the impression that read replicas will pret...

lambrospetrou•4/10/25, 8:21 PM

Unless there is an incident the normal replication lag is a round trip from the primary to the replicas, yes.

Llambrospetrou Unless there is an incident the normal replication lag is a round trip from the ...

ajgeiss0702OP•4/10/25, 8:21 PM

cool, thank you

Samantas5855•4/10/25, 8:22 PM

Ive been facing issues, not all my api pushes make it to the database

Samantas5855•4/10/25, 8:22 PM

I assume Im not the only one

Zack•4/10/25, 8:23 PM

{ "errors": [ { "code": 7500, "message": "internal error; reference = e_28XXLS_d2cd6b5618884603b87fab149025c82e" } ], "success": false, "messages": [], "result": null }

this chart on the blog post gives me the impression that read replicas will pretty much always be un

Similar Threads

Similar Threads

Similar Threads