Hey, so thought I would just raise this here because I have no idea what is happening. We have a dev

Hey, so thought I would just raise this here because I have no idea what is happening. We have a dev D1 database that we deployed on a branch via terraform to test out some stuff. Our main branch ran some other terraform 2 weeks ago and deleted this dev D1 database, however a dev worker that was connected to this D1 database can still read this fine even today. I didn't even realise anything was wrong until I went to run new terraform today on this branch and it complained that the database was missing and needed to be re-created. Not sure what's going on here but feels weird (even though maybe nice for some people if they make a mistake) for the database to just sit around for 2 weeks when it shows as deleted in the dashboard. Who knows how long it''ll actually live on for. The database ID is/was

04a58760-3751-4337-bb3f-29a671a98cdd

04a58760-3751-4337-bb3f-29a671a98cdd

and the worker still connected to it is

tiles-dev

tiles-dev

in our account, specifically deployment

76803c37

76803c37

ZZack Hey, so thought I would just raise this here because I have no idea what is happ...

lambrospetrou•9/3/25, 12:45 PM

Your database is indeed marked deleted in our database (API) (on 2025-08-19 11:51:40.308558+00). When at that point of the delete request we cannot communicate with the DB itself, we have an async flow for the DB (DO) itself to mark itself as deleted on the next heartbeat. I am looking into why that part failed.

Dean•9/4/25, 2:59 AM

is DB performance still impacted in WNAM? seeing pretty slow queries although the query latency metrics looks normal.

Dean•9/4/25, 3:01 AM

"wallTime": 13142 :NotLikeThis:

Dean•9/4/25, 3:22 AM

and magically fixed /shrug

maguayo•9/4/25, 9:38 AM

Hello! I tried D1 like 8 months ago (I think it was in beta). In the website it was anounced as "distributed" or that you could have read replicas in other countries, but in reality that feature wasn't there yet, is it there now? anyone have this information?

Mmaguayo Hello! I tried D1 like 8 months ago (I think it was in beta). In the website it ...

cloudkite•9/4/25, 9:44 AM

https://developers.cloudflare.com/d1/best-practices/read-replication/

Cloudflare Docs

Global read replication

D1 read replication can lower latency for read queries and scale read throughput by adding read-only database copies, called read replicas, across regions globally closer to clients.

Ccloudkite https://developers.cloudflare.com/d1/best-practices/read-replication/

maguayo•9/4/25, 9:45 AM

some of this documentation existed before it was posible to do it, that's why I'm asking for a confirmation if someone is using them

Mmaguayo some of this documentation existed before it was posible to do it, that's why I'...

cloudkite•9/4/25, 9:48 AM

From memory I don’t think there was docs prior to release, but there was a blog post outlining plans and approach

Mmaguayo Hello! I tried D1 like 8 months ago (I think it was in beta). In the website it ...

cloudkite•9/4/25, 9:50 AM

Depending on what you are building durable object SQLite might work better

Mmaguayo some of this documentation existed before it was posible to do it, that's why I'...

lambrospetrou•9/4/25, 12:33 PM

This documentation did not exist before we launched read replication. You are probably remembering something else.
Read replication is supported in D1 for a few months now, in beta yet until we fully implement some nice improvements around replica placement.

What is your use-case?

Llambrospetrou This documentation did not exist before we launched read replication. You are pr...

Larry•9/4/25, 1:31 PM

It's not docs, but this post from 2022 which has this line, "For D1 today, we... dynamically manage the number and location of read-only replicas..." is on the 1st page results when I Google "read-only replicas Cloudflare D1". [https://blog.cloudflare.com/whats-new-with-d1/]

LLarry It's not docs, but this post from 2022 which has this line, "For D1 today, we......

lambrospetrou•9/4/25, 2:57 PM

I think that was for the alpha version of D1, which had completely different architecture from D1 today. Will ping a team internally to see if we can somehow promote the new read replication stuff to surface higher in SEO.

Although googling

d1 read replication

d1 read replication

shows the right material in my own tabs and incognito right at the top.

Llambrospetrou I think that was for the alpha version of D1, which had completely different arc...

lambrospetrou•9/4/25, 3:01 PM

Actually, also

read-only replicas Cloudflare D1

read-only replicas Cloudflare D1

shows the new posts as well.

@Larry where did you see the old post surfacing?

Llambrospetrou Actually, also `read-only replicas Cloudflare D1` shows the new posts as well. ...

Larry•9/4/25, 3:22 PM

When I go incognito, it's not on the 1st page so the algorithm must take into account something about me. I wasn't trying to catch anyone out, I was just trying to explain where the confusion might be coming from. An easier fix for that one page than SEO might be to add an update notice to the blog post itself.

LLarry When I go incognito, it's not on the 1st page so the algorithm must take into ac...

lambrospetrou•9/4/25, 4:12 PM

Yes I already started the banner/notice discussion but I was wondering about the search since I didn't get it

aza547•9/5/25, 7:53 AM

Hello, when I hit the "Enable read replication" button in the dashboard nothing happens. It pulses for a moment then re-appears.

Aaza547 Hello, when I hit the "Enable read replication" button in the dashboard nothing ...

lambrospetrou•9/5/25, 12:53 PM

Do you get any error banner? (jeez, that yellow is rough

)

aza547•9/5/25, 12:57 PM

No error banner I'm afraid

Aaza547 No error banner I'm afraid

lambrospetrou•9/5/25, 1:14 PM

So you still have issues with it? Just saying the obvious, but did you make sure to refresh in case the dashboard accumulated some weird state?
Otherwise, if you still get noop clicks DM me your account and database IDs and i can take a look.

aza547•9/5/25, 1:15 PM

yeah still the same, will DM you

Bbren.app one bit of feedback I've noticed about the D1 Studio! I have a bunch of tables w...

Visal .In•9/5/25, 1:47 PM

The fix lands in production now. Let me know if you still encounter any other problem

Llambrospetrou So you still have issues with it? Just saying the obvious, but did you make sure...

lambrospetrou•9/5/25, 2:00 PM

Just to circle back on that. It's not a read replication issue, there is a regression in our API that returns always "disabled" for read replication mode hence the UI always shows that.
Read replication is still enabled though, and works fine.
We are making a fix now to the API and hopefully will be out by end of the day.

OOr We're seeing an increase of the following error: "D1 DB is overloaded. Too many ...

lambrospetrou•9/5/25, 5:49 PM

https://developers.cloudflare.com/d1/observability/debug-d1/#error-list

That means you send more requests than what your DB can process. It's not a fixed number limit since it depends on your queries. D1 is single threaded, so the slower the queries, the less amount it can execute. Some users do more than 1K/sec, others can only do 10/s.

Anes•9/5/25, 6:38 PM

hey, is there a plan to add transaction support to D1?

OOr Thank you, the queries are quite simple there are just a lot of them due to high...

lambrospetrou•9/5/25, 6:41 PM

It's single threaded, and in each query we return its execution time, so you can track the latency of your queries and judge how many you can handle per second on average based on your traffic.
For example, if you see that your queries are on average 2ms, then you can handle roughly 500.

AAnes hey, is there a plan to add transaction support to D1?

lambrospetrou•9/5/25, 6:41 PM

You mean interleaving with JS code? If yes, not in the short-term.

Llambrospetrou Just to circle back on that. It's not a read replication issue, there is a regre...

lambrospetrou•9/5/25, 6:42 PM

Apologies for this bug that slipped through. The fix should be out to everyone by now, so the UI and the REST API returns the correct

read_replication_mode

read_replication_mode

again.

deternal•9/5/25, 8:23 PM

Hello team, for the past two weeks there were several thousands of D1 errors like network lost/d1 internal errors etc., that happens intermittently So we have tried deleting the data in our D1 database (previously around 4GB) and implemented retries, but now seeing the same level of intermittent errors saying "Exceeded maximum DB size" even the current storage is less than 50MB.

deternal•9/5/25, 8:24 PM

@lambrospetrou Any help would be appreciated, we tried looking at all forums and dont see a solution to this issue. This started only in last 15 days, and impacting our production losing the data

Ddeternal Hello team, for the past two weeks there were several thousands of D1 errors lik...

lambrospetrou•9/5/25, 11:45 PM

Exceeded maximum DB size

This error means that either you are inserting more than the limit for a single DB according to your plan (500MB/10GB), OR, that you are exceeding your account storage limit, again according to your plan.
If you go through support or you account manager they can raise the account limits. The DB size limit cannot be raised higher than 10GB.

VVisal .In The fix lands in production now. Let me know if you still encounter any other pr...

bren.app•9/6/25, 1:46 AM

confirmed it works! Thanks for your efforts here!

aza547•9/6/25, 11:03 AM

After enabling read replication on my D1 DB when should I expect to see the instance count increase from 1? The docs state that one will be spun up in every region, but I'm not seeing that yet (it was only 15 mins ago I enabled it). EDIT: still at 1 instance many hours later.

aza547•9/6/25, 11:06 AM

- Is there something that prevents replica creation if the DB isn't getting enough traffic? My database is very quiet just now (but not silent, it's still receiving queries from all over). But I was still hoping to see the replicas in action.
- Also curious if relicas are spun up on demand within the same region if traffic is high? Didn't see a mention of that in the docs.
I'm trying to dig myself out of a hole where my user database is overloaded during peak times by enabling read replication on it, I'd love to stick with D1 rather than going to another (multi-threaded) offering.

Tux•9/6/25, 1:30 PM

Hello, I am having issues with my D1 database, it is currently down for me. Where do I contact support?

Tux•9/6/25, 1:46 PM

Ah I see the issue. I tried exporting my db that had a virtual table which is now causing the db to not accept any queries

Tux•9/6/25, 1:46 PM

Is there anyway I can force restart my db?

Tux•9/6/25, 3:06 PM

^ Back after 1 hour.

Aaza547 - Is there something that prevents replica creation if the DB isn't getting enou...

lambrospetrou•9/7/25, 8:15 AM

Read replicas have the same lifecycle as a primary DB. If there is no constant traffic for 1-2 minutes, they will hibernate. Once the next request comes in they will start their boot process, and until they finish booting requests are forwarded to the primary. Once the replica boots (depends on the DB size how many (milli)seconds it will take), then replicas serve the read traffic in that region.

If your traffic is very low, once every few minutes for example, then you probably won't see any traffic served by replicas.

My database is very quiet just now (but not silent, it's still receiving queries from all over).

This is not concrete. What is quiet, and what is "still receiving queries from all over"?

Is there something that prevents replica creation if the DB isn't getting enough traffic?

No, replicas are always "created" as in ready to start, one in each region. But, the traffic is what keeps them alive. No traffic, and they will shutdown again.

Llambrospetrou Read replicas have the same lifecycle as a primary DB. If there is no constant t...

lambrospetrou•9/7/25, 8:18 AM

Each response has fields telling you if it's the primary or a replica handling your request and from which region: https://developers.cloudflare.com/d1/best-practices/read-replication/#check-where-d1-request-was-processed

const result = await env.DB.withSession()
  .prepare(`SELECT * FROM Customers WHERE CompanyName = 'Bs Beverages'`)
  .run();
console.log({
  servedByRegion: result.meta.served_by_region ?? "",
  servedByPrimary: result.meta.served_by_primary ?? "",
});

const result = await env.DB.withSession()
  .prepare(`SELECT * FROM Customers WHERE CompanyName = 'Bs Beverages'`)
  .run();
console.log({
  servedByRegion: result.meta.served_by_region ?? "",
  servedByPrimary: result.meta.served_by_primary ?? "",
});

Llambrospetrou Each response has fields telling you if it's the primary or a replica handling y...

lambrospetrou•9/7/25, 8:18 AM

Also, I assume you use the Sessions API. Otherwise, all your requests will go to the primary regardless if you have enabled read replication or not.

Llambrospetrou Also, I assume you use the [Sessions API](https://developers.cloudflare.com/d1/b...

aza547•9/7/25, 8:44 AM

Thank you for the detailed response. That clears it up for me

- It would be awesome if this could be captured in the docs.
- My problem is almost certainly that I'm not using the sessions API.

This is not concrete. What is quiet, and what is "still receiving queries from all over"?

What I was trying to convey is that there is consistent non-zero traffic in every region, and that I expected read replicas in every region to be queried. "Quiet" as in "not enough to overload the primary DB".

The docs to me imply that the sessions API is to guarantee sequential consistency for a set of queries, but the bulk of my traffic is pulling a single field out of the database so I don't actually care about sequential consistency in a lot of cases.

Feel free to point out if I'm missing the obvious, but I don't think the docs make it clear that the sessions API is required for queries to hit a read replica.

I'll go add some

.withSession()

.withSession()

calls to my worker and see what happens. Thanks again.

Aaza547 Thank you for the detailed response. That clears it up for me 🙂 - It would be a...

lambrospetrou•9/7/25, 8:45 AM

At the very top:

Your application can use read replicas with D1 Sessions API.

Your application can use read replicas with D1 Sessions API.

We can make it even stronger if that's not clear enough. I will add another clarifying sentence there.

aza547•9/7/25, 8:50 AM

That would be great, I guess I'm still reading that as "you can use the sessions API optionally to get sequential consistency which a read replication enabled DB otherwise doesn't provide"

aza547•9/7/25, 9:13 AM

Happy days

OOr Will enabling read replication cause the primary DB to be locked while the repli...

lambrospetrou•9/7/25, 9:42 PM

No, replication is done asynchronously, and there is no blocking time while replicas are being created. As I mentioned above though, in order to get benefits from replicas you would need some continuous traffic, especially since 3GB would need a few seconds to boot.
If you only have a request per minute for example, you will pretty much always go to the primary.

const result = await env.DB.withSession() .prepare(`SELECT * FROM Customers WHERE CompanyName = 'Bs Beverages'`) .run(); console.log({ servedByRegion: result.meta.served_by_region ?? "", servedByPrimary: result.meta.served_by_primary ?? "", });

Hey, so thought I would just raise this here because I have no idea what is happening. We have a dev

Similar Threads

Similar Threads

Similar Threads