Skip to content

docs(enterprise): correct the compaction-cleanup-wait default and tuning example - #7869

Merged
jstirnaman merged 2 commits into
masterfrom
pk/compaction-cleanup-wait-default
Oct 9, 2026
Merged

jstirnaman merged 2 commits into
masterfrom
pk/compaction-cleanup-wait-default

Conversation

@praveen-influx

@praveen-influx praveen-influx commented Oct 9, 2026 •

Copy link
Copy Markdown
Contributor

Summary

The configuration reference says the default for --compaction-cleanup-wait is 10m. InfluxDB 3 Enterprise has used 1h since v3.4.2. This PR changes the documented default to 1h. It also removes --compaction-cleanup-wait=10m from the clustering guide's compaction tuning example, because the cleanup wait does not tune compaction.

What changed

  • Configuration options, compaction-cleanup-wait: Default is now 1h instead of 10m.
  • Clustering guide, "Tune compaction parameters": the example no longer sets --compaction-cleanup-wait=10m. The cleanup wait does not change how compaction runs or what it costs; it only sets how long replaced files are kept. Showing it in a tuning example, at a sixth of the default, makes the failure described below more likely.

Why

The default changed from 10m to 1h in v3.4.2 (influxdata/influxdb_pro#1300). The docs kept the old value.

This wait is used by compactor on how long it waits before the compactor can clear the older (older than this wait time) compacted files. If a query node falls behind the configured wait time, its queries can fail with "file not found" errors as the querier could have loaded an older compactor's metadata file pointing to references of data/index files that have been deleted already by compactor.

Impact

Documentation only. Readers who copy the documented default will now get the value the product actually uses.

Verification

  • Checked the default_value of --compaction-cleanup-wait in the influxdb_pro source at every release line (table below).
  • Commit hooks are not installed in my checkout, and I did not run npx hugo --quiet locally. Relying on CI.
Default by release
Release Default
v3.0.3, v3.1.0, v3.2.1, v3.3.0, v3.4.0, v3.4.1 10m
v3.4.2, v3.5.0, v3.6.0, v3.7.0, v3.8.4, v3.9.13, v3.10.6, v3.11.6, v3.12.0, main 1h

The same change moved --catalog-sync-interval from 10s to 1s. The docs already show 1s for that option, so this PR does not touch it.

Preview pages

  • /influxdb3/enterprise/reference/config-options/#compaction-cleanup-wait
    Expected: Default: 1h

  • /influxdb3/enterprise/admin/clustering/#tune-compaction-parameters
    Expected: the example sets --compaction-multipliers and --compaction-max-num-files-per-plan only

Checklist

What changed: the documented default of `--compaction-cleanup-wait` is now `1h` instead of `10m`.

Why: InfluxDB 3 Enterprise changed the default from `10m` to `1h` in v3.4.2 (influxdata/influxdb_pro#1300), and every release since uses `1h`. The docs still showed the old value, so anyone who copies the documented default runs with a much shorter wait than the product uses. The wait is the time query nodes have to move to newly compacted files before the compactor deletes the files they replace.

Impact: documentation only.

Verification: checked the `default_value` of `--compaction-cleanup-wait` in influxdb_pro at v3.4.2, v3.5.0 to v3.12.0 and main (`1h`), and at v3.0.3 to v3.4.1 (`10m`).
@praveen-influx praveen-influx added product:v3-monolith InfluxDB 3 Core and Enterprise (single-node / clusterable) product:shared Shared content across products labels Oct 9, 2026
@github-actions

github-actions Bot commented Oct 9, 2026 •

Copy link
Copy Markdown
Contributor

Vale Style Check Results

Metric Count
Errors 0
Warnings 1
Warnings (1)
File Line Rule Message
content/influxdb3/enterprise/admin/clustering.md 5 InfluxDataDocs.WordList Use 'import' or 'load' instead of 'ingest'.

✅ Check passed

@github-actions

github-actions Bot commented Oct 9, 2026 •

Copy link
Copy Markdown
Contributor

🔗 Link Check Results — Link Check Bot

✅ All links are valid

Metric Value
Files Checked 3
Total Links 1802
Errors 0
Warnings 3
Success Rate 99.50056%
⚠️ 3 warning(s) (do not fail CI)
Source File URL Issue
content/influxdb3/core/reference/config-options/_index.md https://support.influxdata.com/ Network error: SSL certificate not trusted. Use --insecure if site is trusted (e…
content/influxdb3/enterprise/admin/clustering/_index.md https://support.influxdata.com/ Network error: SSL certificate not trusted. Use --insecure if site is trusted (e…
content/influxdb3/enterprise/reference/config-options/_index.md https://support.influxdata.com/ Network error: SSL certificate not trusted. Use --insecure if site is trusted (e…

Full details: workflow run summary and artifact. Last updated: 2026-10-09 10:31:15 UTC

…ning example

What changed: the "Tune compaction parameters" example in the clustering guide no longer sets `--compaction-cleanup-wait=10m`.

Why: the cleanup wait does not change how compaction runs or what it costs. It only sets how long the compactor keeps the files a compaction replaced before it deletes them. A shorter wait gives query nodes less time to move to the new files, so a query node that falls behind can fail queries with "file not found" errors. It does not belong in an example about balancing compaction performance and resource usage, and `10m` is a sixth of the `1h` default.

Impact: documentation only.

Verification: the example's last line no longer ends with a line continuation.
@praveen-influx praveen-influx changed the title docs(enterprise): set compaction-cleanup-wait default to 1h docs(enterprise): correct the compaction-cleanup-wait default and tuning example Oct 9, 2026
--compaction-multipliers=4,8,16 \
--compaction-max-num-files-per-plan=100 \
--compaction-cleanup-wait=10m
--compaction-max-num-files-per-plan=100

Copy link
Copy Markdown
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I'm not sure if the --compaction-cleanup-wait=10m is necessary here given 10m is the default prior to this PR. So, I just removed it as I think reducing this to 10m has side effects on query node as mentioned in the PR description.

@praveen-influx
praveen-influx marked this pull request as ready for review October 9, 2026 10:33
@praveen-influx
praveen-influx requested a review from a team as a code owner October 9, 2026 10:33
@praveen-influx
praveen-influx requested review from jstirnaman and removed request for a team October 9, 2026 10:33
@jstirnaman
jstirnaman added this pull request to the merge queue Oct 9, 2026
Merged via the queue into master with commit 6995712 Oct 9, 2026
36 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

product:shared Shared content across products product:v3-monolith InfluxDB 3 Core and Enterprise (single-node / clusterable)

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants