docs: storage tasks by TC-MO · Pull Request #2975 · apify/apify-docs

TC-MO · 2026-09-14T10:24:29Z

Sharing and cross-run usage were documented three times over, once per storage type, and had drifted apart; this consolidates them into two shared pages (Share storage, Use storage from another run) that the type pages now link to. Along the way it corrects the storage rate limits against apify-core.

The sharing procedure and the between-runs procedure were each written out four times, on the storage overview and all three type pages, with only the noun changed.

The overview's exception list omitted the request queue lock tier and the 40/s batch tier, so it understated the real limits.

The API reference link pointed at key-value stores, the #sharing anchor had no matching heading, and the description promised sharing content.

Record CRUD is a 200/s tier per openapi.yaml, not the 60/s default.

Nesting it under sharing implied a subset relationship that isn't there: granting a person access and addressing your own storage from code are unrelated mechanisms that only shared the word "share".

Request queue limits were lowered in apify-core#29428 (Jul 2026) to avoid DynamoDB partition throttling and the docs were never updated. Also documents the key-value store keys and records list tiers.

Drops a fabricated redoc anchor and corrects the key-value store records endpoint, which downloads a ZIP archive rather than listing records. Restores the concurrent write-order caveat to the storage overview.

The three per-type sections differed only in one method name and one client accessor, so the extraction had relocated duplication rather than removing it. One example per surface plus a lookup table covers all three.

The API client section led with the SDK's "open" flow; the replacement describes the client accessor, and the username/name form it documents is what both clients' to_safe_id conversions accept.

apify-service-account · 2026-09-14T10:25:14Z

✅ Preview for this PR (commit dce12945) is ready at https://pr-2975.preview.docs.apify.com (see action run).

…ence

… anchors Pre-signed URLs cover only selected dataset and key-value store endpoints, so the request queue stub dropped the claim. Headings keep master's #share id; the between-runs fragments decay deliberately - Docusaurus offsets only heading anchors, so span targets hide under the navbar.

One designed SVG, dead links stripped, SVGO) covers what datasets-app.png and find-store-id.png showed. overview-api.png was referenced nowhere.

Restatements only, checked by an independent fidelity review; the one factual change is Python SDK 3.8 -> 3.10, per its pyproject. Also drops the hub's dangling share paragraph and its stale concurrency note.

Fix broken example URLs and leftover filler.

janbuchar

Changes to the OpenAPI spec are fine.

szaganek

Just a few suggestions 🥲 I think we should add a note on using contractions to the style guide.

szaganek · 2026-09-17T12:22:22Z

 Dataset storage enables you to sequentially save and retrieve data. A unique dataset is automatically created and assigned to each Actor run when the first item is stored.

-Typically, datasets comprise results from web scraping, crawling, and data processing jobs. You can visualize this data in a table, where each object is forming a row and its attributes are represented as columns. You have the option to export data in various formats, including JSON, CSV, XML, Excel, HTML Table, RSS or JSONL.
+Typically, datasets hold results from web scraping, crawling, and data processing jobs. The data displays as a table, where each object forms a row and its attributes form columns. You can export the data in JSON, CSV, XML, Excel, HTML table, RSS, or JSONL formats.


Suggested change

Typically, datasets hold results from web scraping, crawling, and data processing jobs. The data displays as a table, where each object forms a row and its attributes form columns. You can export the data in JSON, CSV, XML, Excel, HTML table, RSS, or JSONL formats.

Typically, datasets hold results from web scraping, crawling, and data processing jobs. The data is displayed as a table, where each object forms a row and its attributes form columns. You can export the data in JSON, CSV, XML, Excel, HTML table, RSS, or JSONL formats.

szaganek · 2026-09-17T12:27:30Z

 In [Apify Console](https://console.apify.com), you can view your datasets in the [Storage](https://console.apify.com/storage) section under the [Datasets](https://console.apify.com/storage?tab=datasets) tab.

-![Datasets in app](../images/datasets-app.png)
+![Datasets in Apify Console](../images/storage-types.svg)


In the output, there's some weird space under the screenshot. Can you check if you maybe exported some extra layers?

szaganek · 2026-09-17T12:28:58Z

 1. Click **Export** to download the data in your preferred format.

-Utilize the **Actions** menu to modify the dataset's name, which also affects its [retention period](/storage#data-retention), and to adjust [access rights](/account/collaboration). The **API** button allows you to explore and test the dataset's [API endpoints](/api/v2/storage-datasets).
+Use the **Actions** menu to rename the dataset (which affects its [retention period](/storage#data-retention)) and to adjust [access rights](/account/collaboration). Select **API** to view and test the dataset's [API endpoints](/api/v2/storage-datasets).


It's a good practice to start with the goal. Let's switch:
To rename the dataset...
To view and test...

szaganek · 2026-09-17T12:29:16Z

-Utilize the **Actions** menu to modify the dataset's name, which also affects its [retention period](/storage#data-retention), and to adjust [access rights](/account/collaboration). The **API** button allows you to explore and test the dataset's [API endpoints](/api/v2/storage-datasets).
+Use the **Actions** menu to rename the dataset (which affects its [retention period](/storage#data-retention)) and to adjust [access rights](/account/collaboration). Select **API** to view and test the dataset's [API endpoints](/api/v2/storage-datasets).

 ![Datasets detail view](../images/datasets-detail.png)


Is this screenshot necessary?

szaganek · 2026-09-17T12:29:38Z

-API data push to a dataset is capped at _400 requests per second_ to avoid overloading the servers.
-
-:::
+Pushes to a dataset are rate-limited; see [Rate limiting](#rate-limiting).


Suggested change

Pushes to a dataset are rate-limited; see [Rate limiting](#rate-limiting).

Pushes to a dataset are rate-limited. For details, see [Rate limiting](#rate-limiting).

szaganek · 2026-09-17T13:05:02Z

+
+Share storages by link using their ID or name. Whether the link works for others depends on your account or resource-level general access setting; learn how link-based access works in [General resource access](/account/collaboration/general-resource-access).
+
+For one-off sharing of individual records or items when access is restricted, generate time-limited pre-signed URLs. These are available for selected dataset and key-value store endpoints only. See [Sharing restricted resources with pre-signed URLs](/account/collaboration/general-resource-access#pre-signed-urls).


Let's turn one-off sharing into simple English.

szaganek · 2026-09-17T13:09:18Z

+import Tabs from '@theme/Tabs';
+import TabItem from '@theme/TabItem';
+
+If you have the storage's _name_ or _ID_, you can access it from any [Actor](../actors/index.mdx) or [task](../actors/running/tasks.md) run. Use the same methods and endpoints you'd use for the current run's storages.


Sounds like it's missing a link to these methods, the topic that explains it, or just a plain list.

szaganek · 2026-09-17T13:10:25Z

+
+[Datasets](./dataset/index.md) and [key-value stores](./key_value_store/index.md) support concurrent use. Multiple Actors or tasks can write to the same dataset or key-value store, and multiple runs can read from them at the same time.
+
+[Request queues](./request_queue.md) only allow multiple runs to add new data. A request queue can be processed by one Actor or task run at a time, unless you use [request locking](./request_queue.md#distributivity) to coordinate multiple runs.


Suggested change

[Request queues](./request_queue.md) only allow multiple runs to add new data. A request queue can be processed by one Actor or task run at a time, unless you use [request locking](./request_queue.md#distributivity) to coordinate multiple runs.

[Request queues](./request_queue.md) only allow multiple runs to add new data. A request queue can be processed by one Actor or task run at a time. However, you can use [request locking](./request_queue.md#distributivity) to coordinate multiple runs.

szaganek · 2026-09-17T13:12:26Z

+
+:::info Accessing restricted storage resources between runs
+
+If a storage resource access is set to **Restricted**, the run from which it's accessed must have explicit access to it. Learn how restricted access works in [General resource access](/account/collaboration/general-resource-access), and how to grant it in [Share storage](./share.md).


Suggested change

If a storage resource access is set to **Restricted**, the run from which it's accessed must have explicit access to it. Learn how restricted access works in [General resource access](/account/collaboration/general-resource-access), and how to grant it in [Share storage](./share.md).

If a storage resource access is set to **Restricted**, the run from which it's accessed must have explicit access to it. Learn [how restricted access works](/account/collaboration/general-resource-access) and [how to grant it](./share.md).

Do we need the titles?

szaganek · 2026-09-17T13:13:21Z

+Only the method name changes with the storage type:
+
+| Storage type    | [JavaScript SDK](/sdk/js)                                                        | [Python SDK](/sdk/python)                                                                 |
+| --------------- | -------------------------------------------------------------------------------- | ----------------------------------------------------------------------------------------- |


How the heck did this table pass the checks, mine always fail if there's as much as a single extra character :D

TC-MO added 10 commits September 4, 2026 22:54

docs: rename dataset validation and multiple datasets to task titles

f4c9f7b

docs: extract storage sharing into two shared topics

2000dd0

The sharing procedure and the between-runs procedure were each written out four times, on the storage overview and all three type pages, with only the noun changed.

docs: route storage rate limiting to per-type tiers

b9c85b4

The overview's exception list omitted the request queue lock tier and the 40/s batch tier, so it understated the real limits.

fix: correct stale links and description on the request queue page

be5a795

The API reference link pointed at key-value stores, the #sharing anchor had no matching heading, and the description promised sharing content.

docs: correct the key-value store rate limit to 200 per second

d12dcf5

Record CRUD is a 200/s tier per openapi.yaml, not the 60/s default.

docs: rename share between runs to use storage from another run

50f0a4e

Nesting it under sharing implied a subset relationship that isn't there: granting a person access and addressing your own storage from code are unrelated mechanisms that only shared the word "share".

docs: correct storage rate limits against apify-core

cdc4fe0

Request queue limits were lowered in apify-core#29428 (Jul 2026) to avoid DynamoDB partition throttling and the docs were never updated. Also documents the key-value store keys and records list tiers.

docs: fix defects found in review of the storage restructure

4aae36f

Drops a fabricated redoc anchor and corrects the key-value store records endpoint, which downloads a ZIP archive rather than listing records. Restores the concurrent write-order caveat to the storage overview.

docs: deduplicate the use-from-another-run samples

8801b8c

The three per-type sections differed only in one method name and one client accessor, so the extraction had relocated duplication rather than removing it. One example per surface plus a lookup table covers all three.

docs: tighten prose on the storage sharing pages

4a33e2d

The API client section led with the SDK's "open" flow; the replacement describes the client accessor, and the username/name form it documents is what both clients' to_safe_id conversions accept.

TC-MO self-assigned this Sep 14, 2026

TC-MO added documentation Improvements or additions to documentation. t-docs Issues owned by technical writing team. labels Sep 14, 2026

TC-MO changed the title ~~Docs/storage tasks~~ docs: storage tasks Sep 14, 2026

TC-MO added 4 commits September 14, 2026 12:37

docs: unstack the admonitions and link the SDK methods to their refer…

11a429e

…ence

docs: replace storage Console screenshots with the storage-types graphic

ea35b8d

One designed SVG, dead links stripped, SVGO) covers what datasets-app.png and find-store-id.png showed. overview-api.png was referenced nowhere.

docs: add the KVS and RQ storage-types graphics

f7f4672

TC-MO marked this pull request as ready for review September 15, 2026 16:30

TC-MO requested review from fnesveda, janbuchar and szaganek as code owners September 15, 2026 16:30

TC-MO added 2 commits September 15, 2026 19:18

docs: cut filler and fix grammar across the storage pages

a70c533

Restatements only, checked by an independent fidelity review; the one factual change is Python SDK 3.8 -> 3.10, per its pyproject. Also drops the hub's dangling share paragraph and its stale concurrency note.

docs: fixes

dce1294

Fix broken example URLs and leftover filler.

janbuchar approved these changes Sep 16, 2026

View reviewed changes

szaganek reviewed Sep 17, 2026

View reviewed changes

Navigation Menu

docs: storage tasks - #2975

docs: storage tasks#2975
TC-MO wants to merge 16 commits into
masterfrom
docs/storage-tasks

TC-MO commented Sep 14, 2026

Uh oh!

apify-service-account commented Sep 14, 2026 •

edited

Loading

Uh oh!

janbuchar left a comment

Uh oh!

szaganek left a comment

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

szaganek Sep 17, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

	Typically, datasets hold results from web scraping, crawling, and data processing jobs. The data displays as a table, where each object forms a row and its attributes form columns. You can export the data in JSON, CSV, XML, Excel, HTML table, RSS, or JSONL formats.
	Typically, datasets hold results from web scraping, crawling, and data processing jobs. The data is displayed as a table, where each object forms a row and its attributes form columns. You can export the data in JSON, CSV, XML, Excel, HTML table, RSS, or JSONL formats.

	Pushes to a dataset are rate-limited; see [Rate limiting](#rate-limiting).
	Pushes to a dataset are rate-limited. For details, see [Rate limiting](#rate-limiting).


		Share storages by link using their ID or name. Whether the link works for others depends on your account or resource-level general access setting; learn how link-based access works in [General resource access](/account/collaboration/general-resource-access).

		For one-off sharing of individual records or items when access is restricted, generate time-limited pre-signed URLs. These are available for selected dataset and key-value store endpoints only. See [Sharing restricted resources with pre-signed URLs](/account/collaboration/general-resource-access#pre-signed-urls).


		[Datasets](./dataset/index.md) and [key-value stores](./key_value_store/index.md) support concurrent use. Multiple Actors or tasks can write to the same dataset or key-value store, and multiple runs can read from them at the same time.

		[Request queues](./request_queue.md) only allow multiple runs to add new data. A request queue can be processed by one Actor or task run at a time, unless you use [request locking](./request_queue.md#distributivity) to coordinate multiple runs.


		:::info Accessing restricted storage resources between runs

		If a storage resource access is set to Restricted, the run from which it's accessed must have explicit access to it. Learn how restricted access works in [General resource access](/account/collaboration/general-resource-access), and how to grant it in [Share storage](./share.md).

Conversation

TC-MO commented Sep 14, 2026

Uh oh!

apify-service-account commented Sep 14, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

janbuchar left a comment

Choose a reason for hiding this comment

Uh oh!

szaganek left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

4 participants

apify-service-account commented Sep 14, 2026 •

edited

Loading