laantungir/fips
1
0
Fork 0
mirror of https://github.com/jmcorgan/fips.git synced 2026-07-22 07:48:26 +00:00
Code Issues Packages Projects Releases Wiki Activity

Make the admission-cap final peer_count read tolerate a busy daemon

Browse Source 
The suite red-ed with an empty final peer_count while every substantive
assertion passed: both denied peers showed inbound handshake attempts and
zero outbound responses, which is the gate this suite exists to verify.

Two pre-existing defects combined. The final count was sampled once, in
the same second the load driver issues its last round of peer restarts, so
the read can land on a daemon busy with those restarts and return nothing.
And the fallback meant to cover that could never fire, because the shell
applies it to the last stage of the pipeline, which succeeds on empty
input - so an unanswered query was indistinguishable from a genuine zero,
and reported as a cap violation.

Poll the final read for a short window instead, and give the two cases
distinct output. A real regression still fails, just after the retry
window, since the loop exits early only on the expected value. Extract the
read the two phases share rather than leaving them to drift.
This commit is contained in:
Johnathan Corgan
2026-07-19 06:46:50 +00:00
parent e4a854f6b0
commit 3ebb14eda4
1 changed files with 27 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

32
testing/static/scripts/admission-cap-test.sh
View File

@@ -87,13 +87,23 @@ CAP_IP=$(node_ip "$CAP_NODE")
[ -n "$CAP_IP" ] || fail "could not resolve docker_ip for node-$CAP_NODE in $TOPO_FILE"
info "cap'd node: node-$CAP_NODE (ip $CAP_IP, max_peers=$MAX_PEERS)"
# Read the cap'd node's peer_count, or the empty string if it did not answer.
#
# Empty is deliberately distinct from a real 0. An `|| echo 0` fallback cannot
# do that job here: `||` binds to the last stage of the pipeline, which succeeds
# on empty input, so the fallback never fires and an unreachable container would
# be reported as a legitimate zero.
read_peer_count() {
docker exec "fips-node-${CAP_NODE}${FIPS_CI_NAME_SUFFIX:-}" fipsctl show status 2>/dev/null \
| grep -m1 peer_count | sed 's/.*: *//' | tr -d ','
}
# ── Phase 1: wait for convergence ────────────────────────────────────
info "phase 1: wait for node-$CAP_NODE peer_count to reach $MAX_PEERS (90s timeout)"
deadline=$(($(date +%s) + 90))
pc=0
while [ "$(date +%s)" -lt "$deadline" ]; do
pc=$(docker exec "fips-node-${CAP_NODE}${FIPS_CI_NAME_SUFFIX:-}" fipsctl show status 2>/dev/null \
| grep -m1 peer_count | sed 's/.*: *//' | tr -d ',' || echo 0)
pc=$(read_peer_count)
[ "$pc" = "$MAX_PEERS" ] && break
sleep 2
done
@@ -186,9 +196,21 @@ for n in $DENIED; do
done
# ── Phase 4: cap'd node still at exactly max_peers ───────────────────
pc_final=$(docker exec "fips-node-${CAP_NODE}${FIPS_CI_NAME_SUFFIX:-}" fipsctl show status 2>/dev/null \
| grep -m1 peer_count | sed 's/.*: *//' | tr -d ',' || echo 0)
info "node-$CAP_NODE final peer_count=$pc_final (expected $MAX_PEERS)"
#
# Polled rather than sampled once. The load driver above restarts the denied
# peers every 15s and its final restart lands in the same second this runs, so a
# single read can hit a daemon busy with those restarts and come back empty --
# which is a harness race, not a cap regression. Retry briefly before believing
# the answer. A genuine regression still fails, just after the retry window,
# because the loop exits early only on the expected value.
pc_final=""
deadline=$(($(date +%s) + 30))
while [ "$(date +%s)" -lt "$deadline" ]; do
pc_final=$(read_peer_count)
[ "$pc_final" = "$MAX_PEERS" ] && break
sleep 2
done
info "node-$CAP_NODE final peer_count=${pc_final:-<no answer>} (expected $MAX_PEERS)"
[ "$pc_final" = "$MAX_PEERS" ] || OVERALL=1
if [ "$OVERALL" -eq 0 ]; then