codenotary / immudb
85%
master: 85%

LAST BUILD BRANCH: chore/test-followups
DEFAULT BRANCH: master
Repo Added
Files 311
Badge
Embed ▾
README BADGES
x

If you need to use a raster PNG badge, change the '.svg' to '.png' in the link

Markdown

Textile

RDoc

HTML

Rst

LAST BUILD ON BRANCH chore/test-followups
branch: chore/test-followups
CHANGE BRANCH
x

coverage: 85.034% (-0.004%) from 85.038%
25365079973

Pull #2095

gh-ci

vchaindz 
chore(tests): add hot-restore reproducer for issue #2061

Drives a two-instance src/dst immudb setup through a SQL workload
designed to surface the "indexing failed at ... due to error: tbtree:
key not found" + 60s freeze symptom reported in #2061. The script:

  - Boots two immudb instances (src on :3322, dst on :3323) with
    deterministic admin passwords and isolated data dirs under
    /tmp/immudb-2061.
  - Builds ~2 GB of payload on the source: 2M INSERTs into a table
    with a UNIQUE composite index on (k1, k2), 200k UPDATEs on the
    composite-key column, plus two extra phases targeting the
    InjectiveMapping / GetBetween+ReadTxEntry path:
      D - multi-statement INSERT-new + UPDATE-other in one exec call
      E - INSERT/DELETE/INSERT cycles reusing composite-key values
          across separate transactions
  - immuadmin hot-backup -> immuadmin hot-restore --force-replica
    into a fresh dst DB.
  - 5-minute observation window grepping dst.log for the failure
    strings, plus post-restore SELECTs through the composite-index
    path so a silent ErrKeyNotFound from the indexer would surface
    at read time.
  - Prints REPRO=yes/no.

Confirmed REPRO=no on master (47336 transactions, 2.1 GB backup,
zero "indexing failed" / "tbtree: key not found" lines, post-restore
SELECTs return expected counts and rows) - most plausibly closed
silently by the indexer/tbtree work shipped since v1.9.7.

Side finding while building this: INSERT + UPDATE on the same row
in a single exec call still trips ErrCannotUpdateKeyTransiency on
master, despite 35bb7962 ("disjoint keyRef spaces") supposedly
covering this class. Filed separately as #2092 with a minimal repro.

Run with:
  bash tests/repro/issue-2061/run.sh
or scaled down for a quick smoke test:
  TOTAL=600 BATCH=60 MULTI_STMT_COUNT=50 CYCLE_COUNT=30 \
    OBSERVE_SECS=10 bash tests/repro/issue-2061/run.sh
Pull Request #2095: test: INSERT+UPDATE same-row regression coverage and issue-2061 hot-restore reproducer

45155 of 53102 relevant lines covered (85.03%)

126342.41 hits per line

Relevant lines Covered
53102 RELEVANT LINES 45155 COVERED LINES
126342.41 HITS PER LINE
Source Files on chore/test-followups

Recent builds

Builds Branch Commit Type Ran Committer Via Coverage
25365079973 chore/test-followups chore(tests): add hot-restore reproducer for issue #2061 Drives a two-instance src/dst immudb setup through a SQL workload designed to surface the "indexing failed at ... due to error: tbtree: key not found" + 60s freeze symptom reported in #2061... Pull #2095 vchaindz gh-ci
85.03
See All Builds (4068)