Monitor: keep checking interfaces whose link-down alert is still open
Lint / JS (eslint) (push) Successful in 8s
Lint / Python (flake8) (push) Successful in 49s
Lint / Notify on failure (push) Skipped
Lint / Deploy (push) Skipped
Security / Python Security (bandit) (push) Successful in 53s
Test / Python Tests (pytest) (push) Successful in 1m3s
Lint / Python (flake8) (pull_request) Successful in 55s
Lint / JS (eslint) (pull_request) Successful in 6s
Lint / Notify on failure (pull_request) Skipped
Lint / Deploy (pull_request) Skipped
Security / Python Security (bandit) (pull_request) Successful in 1m22s
Test / Python Tests (pytest) (pull_request) Successful in 1m27s
Lint / JS (eslint) (push) Successful in 8s
Lint / Python (flake8) (push) Successful in 49s
Lint / Notify on failure (push) Skipped
Lint / Deploy (push) Skipped
Security / Python Security (bandit) (push) Successful in 53s
Test / Python Tests (pytest) (push) Successful in 1m3s
Lint / Python (flake8) (pull_request) Successful in 55s
Lint / JS (eslint) (pull_request) Successful in 6s
Lint / Notify on failure (pull_request) Skipped
Lint / Deploy (pull_request) Skipped
Security / Python Security (bandit) (pull_request) Successful in 1m22s
Test / Python Tests (pytest) (pull_request) Successful in 1m27s
After the interface baseline is reset, a port that is already down reads as "first seen down", which neither alerts nor resolves. An alert opened before the reset was then stuck open without updates: micro1's enp3s0 had sat at one failed check since March with no ticket. An interface with an open interface_down alert is now treated as previously up, so the alert keeps counting, tickets and resolves as normal. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
@@ -1195,6 +1195,10 @@ class NetworkMonitor:
|
||||
def _process_interfaces(self, states: Dict[str, Dict[str, bool]], suppressions: list) -> None:
|
||||
baseline = db.get_baseline()
|
||||
new_baseline = {k: dict(v) for k, v in baseline.items()}
|
||||
# An interface with an open link-down alert was up before it went down, whatever the
|
||||
# baseline says now: after a baseline reset it reads as "first seen down", which would
|
||||
# leave the alert open forever without updating, ticketing or resolving it.
|
||||
open_down = set(db.get_open_event_targets('interface_down'))
|
||||
# Only count hosts with genuine regressions (UP→DOWN) toward cluster threshold
|
||||
hosts_with_regression: List[str] = []
|
||||
|
||||
@@ -1207,6 +1211,8 @@ class NetworkMonitor:
|
||||
|
||||
for iface, is_up in ifaces.items():
|
||||
prev = baseline.get(host, {}).get(iface) # 'up', 'initial_down', or None
|
||||
if (host, iface) in open_down:
|
||||
prev = 'up'
|
||||
|
||||
if is_up:
|
||||
new_baseline[host][iface] = 'up'
|
||||
|
||||
Reference in New Issue
Block a user