Retry a failed blob upload five times, 3 seconds apart (closes #476) #479

Open
clawbot wants to merge 1 commits from issue-476-retry-blob-upload into next
pull from: issue-476-retry-blob-upload
Collaborator

closes #476

A failed blob upload now gets up to five retries, 3 seconds apart, before the backup stops with the upload's error.

  • Each retry rewinds the blob's temporary file and uploads the whole blob again. No backend closes the reader it is given, so the rewind works on S3, rclone and file destinations.
  • A warning is printed before each retry. It counts toward the "Finished with N warning(s)" line, so --cron reports a backup that needed a retry.
  • An interrupt during an upload or during the wait stops at once, without a retry.
  • The panic in the report ("close of closed channel" in the progress reporter's Stop) was already fixed on next by the change for #253. A new test runs a backup without --cron whose blob uploads keep failing and checks that it returns the upload error.

Disclosures:

  • Judgement call: "retry by default 5 times" is read as five retries after the first try, so six tries in all. The count and the delay are constants, not config keys. Adding a config key later only means adding the key, its check and its README entry.
  • The delay is a ScannerConfig field. Production sets it to 3 seconds; tests leave it zero so the existing fault-injection tests stay fast.

Model: opus-5-5

closes https://git.eeqj.de/sneak/vaultik/issues/476 A failed blob upload now gets up to five retries, 3 seconds apart, before the backup stops with the upload's error. - Each retry rewinds the blob's temporary file and uploads the whole blob again. No backend closes the reader it is given, so the rewind works on S3, rclone and file destinations. - A warning is printed before each retry. It counts toward the "Finished with N warning(s)" line, so `--cron` reports a backup that needed a retry. - An interrupt during an upload or during the wait stops at once, without a retry. - The panic in the report ("close of closed channel" in the progress reporter's `Stop`) was already fixed on `next` by the change for https://git.eeqj.de/sneak/vaultik/issues/253. A new test runs a backup without `--cron` whose blob uploads keep failing and checks that it returns the upload error. Disclosures: - Judgement call: "retry by default 5 times" is read as five retries after the first try, so six tries in all. The count and the delay are constants, not config keys. Adding a config key later only means adding the key, its check and its README entry. - The delay is a `ScannerConfig` field. Production sets it to 3 seconds; tests leave it zero so the existing fault-injection tests stay fast. Model: opus-5-5
clawbot added the needs-review label 2026-11-19 10:29:00 +01:00
clawbot self-assigned this 2026-11-19 10:29:00 +01:00
clawbot added 1 commit 2026-11-19 10:29:00 +01:00
A blob upload that failed, such as one cut off by a lost connection,
ended the backup at once. It is now tried again from the start of the
blob up to five times, 3 seconds apart, with a warning before each
retry; an interrupt still stops it at once. The panic in the report,
"close of closed channel" from the progress reporter, was already fixed
on next by the commit for issue 253; a test now covers a failing upload
with the progress reporter on.

Model: opus-5-5
clawbot added the needs-decision label 2026-11-19 10:39:08 +01:00
Author
Collaborator

Verdict: PASS
Head: d71be0b4c8

A failed blob upload is now retried five times, 3 seconds apart, from the start of the blob on every backend, and the backup stops with the upload's error after the last try.

  • Judgement call: sneak's "retry by default 5 times with 3 second delays" is read, as the PR body reads it, as fixed values with no config key.
  • Open point for sneak: he closed #476 after saying the report may come from a very old version; this review takes that issue's retry request as still wanted.

Model: opus-5-5

Verdict: PASS Head: d71be0b4c890da9ebf5402bc8fd48707e91d3968 A failed blob upload is now retried five times, 3 seconds apart, from the start of the blob on every backend, and the backup stops with the upload's error after the last try. - Judgement call: sneak's "retry by default 5 times with 3 second delays" is read, as the PR body reads it, as fixed values with no config key. - Open point for sneak: he closed https://git.eeqj.de/sneak/vaultik/issues/476 after saying the report may come from a very old version; this review takes that issue's retry request as still wanted. Model: opus-5-5
Author
Collaborator

Held for sneak: he closed the issue after noting the report might be from an old version. The panic is already fixed on next; this PR adds the retries the issue asked for. It passed review. Merge or close is his call.

Model: opus-5-5

Held for sneak: he closed the issue after noting the report might be from an old version. The panic is already fixed on `next`; this PR adds the retries the issue asked for. It passed review. Merge or close is his call. Model: opus-5-5
clawbot removed their assignment 2026-11-19 11:09:19 +01:00
sneak was assigned by clawbot 2026-11-19 11:09:19 +01:00
Some checks are pending
check / check (push) Waiting to run
You are not authorized to merge this pull request.
This pull request can be merged automatically.
This branch is out-of-date with the base branch
View command line instructions

Checkout

From your project repository, check out a new branch and test the changes.
git fetch -u origin issue-476-retry-blob-upload:issue-476-retry-blob-upload
git checkout issue-476-retry-blob-upload
Sign in to join this conversation.