Coinbase Details 50-Minute Outage Caused by Configuration Error

Coinbase Says a Name Collision Cut Off Inbound Traffic

Coinbase said it deployed a routine configuration change at 12:34 p.m. ET as part of a migration to a new service-deployment model. The company described the change as low risk and said it passed its normal review and deployment process.

The update inadvertently modified Kubernetes resources associated with an Istio ingress gateway because the resources had colliding names that pre-production checks did not detect, Coinbase said. An ingress gateway regulates traffic entering the cluster, and Coinbase said the resulting failure stopped all inbound traffic by 12:37 p.m. ET.

That traffic loss reached nearly all internal asynchronous workflows, according to the company. In simpler terms, systems that queue and complete operational steps in the background could no longer reach the services they needed, so transactions and settlement activity paused rather than proceeding normally.

Coinbase said the affected gateway was restored at 1:20 p.m. ET and the incident was mitigated at 1:23 p.m. ET. Queued workflows then completed, with most services caught up shortly afterward and remaining backlogs clearing over the following hours.

The direct technical fix was to redeploy the gateway at its latest stable version. Coinbase said the recovery was slowed by a circular dependency: the deployment tooling engineers ordinarily use relied on the same gateway that was unavailable.

The company said engineers instead invoked a rollback workflow manually through its cloud provider using just-in-time privileged access. That emergency route worked, Coinbase said, but its permission checks and procedural controls added time during the recovery.

Card Payments, Transfers and Onchain Swaps Were Disrupted

The customer impact went beyond a trading screen. Coinbase said retail users were unable to complete off-platform trades, deposits and withdrawals during the incident, while in-flight transactions appeared stuck rather than failed and completed after service returned.

Coinbase Card debit transactions were declined for the duration, the company said. Credit-card purchases continued, but card-management features were unavailable, a distinction that narrows the scope of the impact but leaves no indication of how many debit attempts were rejected.

Coinbase also said swaps through Coinbase DEX on Base and Solana were unavailable. Its Exchange and Prime customers saw failed or delayed transfers and settlement activity, while Developer Platform users could not onboard, move funds or use onramp services.

Those different products share a practical dependency even though they serve different customers. Daily Crypto Briefs recently examined Coinbase’s push to put USDC into regulated collateral workflows through its Marex margin-clearing integration; the outage illustrates that expanding that product stack also expands the operational surface that must remain available.

Coinbase said it halted affected flows and posted updates on its Retail, Exchange and Prime status pages while recovery was underway. Its current status page lists the principal website, mobile, advanced-trade and API services as operational, but a current dashboard does not measure the July 14 backlog or individual customer outcomes.

The company said no customer funds were at risk, a statement about custody and asset safety rather than a claim that every user experienced no interruption. It directed customers with persistent issues from the event to its support channels.

Coinbase Plans New Deployment Guardrails After the Outage

Coinbase said it is extending deploy-time guardrails to detect and block Kubernetes resource-name collisions, and is stress-testing those checks against other plausible failures. The stated aim is to stop a deployment from modifying resources owned by neighboring components.

It also said it will improve redundancy between deployment tooling and the infrastructure that tooling manages. The goal is to avoid a repeat of the recovery constraint in which the normal tool needed to repair an outage becomes impaired by the outage itself.

The third action is to audit and exercise its break-glass access paths, Coinbase said. Such paths are intentionally more restrictive than ordinary access, but the company said it will work to make the manual recovery route faster and reliable when it is required.

For active market participants, the relevant evidence will be operational rather than promotional: whether comparable incidents recur, how quickly affected workflows clear and whether Coinbase later publishes measures for transaction impact or recovery time. The company has not given a date for the guardrail changes or a public service-level target tied to this incident.

The broader industry is also concentrating more activity inside venue infrastructure. Coinbase’s planned unification of its derivatives technology with Deribit, described in Daily Crypto Briefs’ coverage of the Coinbase-Deribit matching-engine release, makes resilience and recoverability increasingly consequential as product lines converge.

Broader crypto sentiment remained cautious. The Crypto Fear & Greed Index read 31, classified as Fear, on July 23, a market-wide gauge that does not measure confidence in Coinbase specifically.

Coinbase’s next test is whether the safeguards it described prevent another configuration change from producing the same failure pattern. The company has established the timeline and root cause, but it has not disclosed affected customer counts, transaction values, the residual backlog by product or a completion date for its remediation work.