Running on Backblaze B2?

Backblaze B2 lifecycle or S3 compat failing? We fix it.

A nightly backup job started returning TooManyBuckets after someone wrote a script that creates a bucket per tenant, lifecycle rules that were supposed to clean up old versions quietly stopped firing, and the S3-compatible endpoint is returning SignatureDoesNotMatch for a region clients copied from the AWS documentation. We open the b2 CLI and the S3 logs side by side, work out which quirk is biting, and get backups restoring cleanly.

  • Since 2015Fixing technology
  • FreeEstimates
  • US teamReal engineers

backblaze-b2.log

still broken

What it is

Backblaze B2 at a glance

Backblaze B2 Cloud Storage is a low-cost object storage service from Backblaze, Inc., the company best known for Computer Backup, their consumer and SMB backup product. B2 launched in 2015 as a direct, deliberately cheap alternative to Amazon S3, with storage and egress priced well below hyperscaler rates and a free egress allowance of three times stored data per month, which makes it a serious target for backups, media archives and content delivery when paired with Cloudflare's Bandwidth Alliance.

B2 exposes two APIs: a native B2 API and an S3-compatible endpoint that behaves like a close cousin of S3 rather than a perfect clone. Most trouble in production clusters around the compatibility edges, the 100-bucket default limit, Application Keys whose scope is narrower than the client expects, lifecycle rules that interact in surprising ways with Object Lock, and large uploads whose part sizes do not match B2's optimal chunking. The service is reliable in the sense that matters, with multi-drive replication and a strong durability record, and it demands a little more reading of the fine print than a hyperscaler of equivalent spend would.

Type
Low-cost object storage with S3-compatible API
First released
2015
License
Proprietary service, open-source SDKs and CLI
Maintained by
Backblaze, Inc.
Typically used for
Backup targets, media archives, static asset hosting behind Cloudflare, cold data retention, disaster recovery stores

The honest version

Backblaze B2 pros and cons

No technology is all upside. This is where Backblaze B2 earns its place and where it tends to cause trouble.

Where Backblaze B2 is strong

  • Storage priced an order of magnitude below the hyperscalersSix dollars per terabyte per month is not a headline number; it is the catalog price. For long-term backup, media archives and static asset buckets the math is obvious.
  • Free egress allowance of three times storageTeams that read their data regularly often pay nothing for egress, which flips the economics of backup restore testing from scary to routine.
  • S3-compatible endpointExisting S3 SDKs, Terraform providers and client tools work against B2 with a different endpoint and credentials, which keeps migrations short rather than a rewrite.
  • Bandwidth Alliance with CloudflareEgress from B2 to Cloudflare's network is free, which makes B2 behind a Cloudflare worker or CDN a very cheap way to serve large media without a dedicated CDN contract.
  • Object Lock and server-side encryptionWORM-style retention through Object Lock and SSE-B2 or SSE-C encryption are first-class features rather than add-ons, which clears the common compliance boxes for backups and archives.
  • Honest status page and supportStatus updates arrive early on incidents, support tickets reach a real engineer, and the company has a long history of writing openly about the hardware behind the service.

Where Backblaze B2 bites

  • Hundred-bucket limit catches people outThe default account cap of 100 buckets feels infinite until the script that creates a bucket per customer runs into TooManyBuckets. Raising the cap is a support request, and rearchitecting around the limit sometimes makes more sense.
  • S3 compatibility is close, not exactMost SDK operations work; some multipart edge cases, certain ACL semantics and a handful of headers behave differently enough to break automation that assumes AWS behavior.
  • Only a few regionsA handful of US and EU regions. Latency-sensitive applications close to Asia-Pacific or South America have to cross the backbone to reach the nearest B2 cluster.
  • Application Keys are strictKeys scoped to a specific bucket or operation will not silently work against a different bucket, which is good for security and surprising for an engineer who copied a key from an older project.
  • Daily download capsAccounts have a configurable daily download cap that blocks further reads once hit. This protects against runaway bills and causes outages when the cap is set too low for a real restore.
  • Limited native managed services around the storeB2 is deliberately just object storage. There is no managed CDN, queue, database or compute in the catalog, so richer stacks live elsewhere and treat B2 as a storage tier.

Sound familiar?

Common Backblaze B2 problems we fix

The Backblaze B2 problems that bring people to us, and what you have once each one is fixed.

too_many_buckets, max is 100

Bucket creation is failing at the account cap

A tenant-per-bucket design that worked for the first eighty customers and stops at the hundredth. Raising the cap helps for a while; the deeper fix is usually a prefix-per-tenant model.

Bucket strategy aligned to real tenancy

SignatureDoesNotMatch

S3 clients cannot sign requests against B2

A region name copied from AWS examples, an endpoint URL that does not match the bucket's region, or a clock skew on the client. B2 is strict about signature V4 details.

Region, endpoint and clock aligned

cap_exceeded: daily download cap

A real restore hit the daily download cap

A disaster recovery drill or an actual restore that needed more data than the account's download cap allowed in one day, so the job stopped halfway through with files missing.

Caps set to real restore sizes

lifecycle rule has not run in 14d

Lifecycle rules do not fire as expected

An Object Lock retention that holds versions past the lifecycle delete, or a rule written against the wrong file prefix, so old versions accumulate and the storage bill grows.

Lifecycle and Object Lock working together

Application Key scoped to 'prod' used against 'staging'

Application Key scope does not match the client

A key created with narrow permissions that is being used from a process that expects wider access, or the other way. The error is honest and often misread as a transient failure.

Keys scoped intentionally and documented

large file upload: part size 5MB

Large uploads are slow or failing

Multipart uploads using the AWS-default 5MB part size against B2, which prefers 100MB parts for large files. Transfers are slow, part counts hit limits, and reassembly occasionally fails.

Part sizes tuned to B2's shape

expired_auth_token

Native B2 clients lose authorization

A long-running process that cached the B2 authorization token past its 24-hour window without re-authorizing, so every subsequent call returns 401 and the job stalls.

Auth refresh handled explicitly

503 service_unavailable on us-west-004

A region returns transient 503s

A regional B2 cluster under temporary pressure, often during maintenance. Clients without a retry and backoff plan surface the error to users or fail the batch.

Retries with backoff in every client

What we do

Backblaze B2 help, from migration to steady running

Ring us for one failing restore, or let us own the whole object-storage tier.

Migration onto B2
Moving backups, media archives or static buckets from S3, GCS or Azure Blob to B2 with rclone, aws s3 sync or the native API, keeping lifecycle intent intact.
Backup and restore design
Buckets, prefixes, Object Lock and lifecycle rules set so backups go in cleanly and restores come out predictably, with a drill schedule to prove it.
Cloudflare and CDN integration
B2 behind Cloudflare or another CDN through the Bandwidth Alliance, with cache rules and signed URLs so media serves cheaply and safely.
S3 compatibility rescue
Clients and SDKs that work on AWS and break on B2 brought back to parity by fixing endpoints, regions, signature settings and multipart configuration.
Cost review
Egress, storage class and bucket count reviewed against the real access pattern, with a before and after bill estimate.
Managed Tech
Lifecycle verification, restore drills, Application Key rotation and someone to page when a daily cap trips mid-restore.

How it works

From first message to fixed

  1. Look Free estimate

    Tell us what's going wrong. We look at it, then tell you plainly what's wrong and what it will cost.

  2. Fix

    We fix the cause, not the symptom. If something is down, getting it back up comes first.

  3. Harden

    Backups, monitoring, updates and access, so the same problem can't quietly come back.

  4. Hand back or stay on

    You get a plain-language write-up. Or we stay on for hosting, upkeep and new work.

Often found together

Running Backblaze B2 alongside something else?

We work on the whole stack, so the problem can be wherever it actually is.

Questions

Backblaze B2 questions

Something else? Call+1 737 727 7327

Should we move our backups from S3 to Backblaze B2?

For long-term retention and infrequently-read backups, usually yes. The storage price and the free egress allowance dominate the economics once data crosses a few terabytes. For hot application data that lives close to compute in a specific AWS region, S3 still wins on latency and on integration with surrounding AWS services. We will model both bills honestly rather than push a single answer.

Why does our S3 SDK work on AWS and fail on B2?

Usually a region or endpoint mismatch. B2 uses region names like us-west-004 or eu-central-003 rather than AWS region names, and the endpoint URL must match. Clock skew on the client also triggers signature errors because B2 is strict about signature V4 timestamps. Certain multipart edge cases differ too, and we work through the client configuration until the compatibility edges stop biting.

How do we handle the 100-bucket limit?

First by asking whether a bucket per tenant is actually the right design. Most production setups work better with one bucket per environment and a prefix per tenant, which scales without hitting any account limit. If a bucket per tenant is genuinely required, Backblaze will raise the cap on request, and we plan for the next limit before the first one bites again.

Can you take over a Backblaze B2 account from a previous team?

Yes. We inventory the buckets, Application Keys and lifecycle rules, rotate keys into your organization, confirm restores actually succeed from the current state, and bring the account structure under Terraform or a documented runbook before any changes land.

How do we restore quickly without blowing the daily download cap?

First by setting a cap that matches the real size of a worst-case restore rather than the day-to-day traffic. We also stage restores through a parallel reader with careful part sizes, use the Cloudflare Bandwidth Alliance when the destination allows, and split very large restores across days if a cap genuinely cannot be raised in time.

How much does Backblaze B2 help cost?

Estimates are free of charge. Send us the bucket layout, the client stack and the behavior you are seeing, and we return a fixed price before any changes touch the account.

Free estimate

Backblaze B2 giving you trouble?

Tell us what you're running and what's going wrong. A person reads it and comes back with an estimate.

  • No charge to find out what it will take
  • If it's urgent, say so

Replies come from a person. By sending, you authorize us to email you back. No marketing.