Commit Graph

14 Commits

Author SHA1 Message Date
andreas 475319e248 fix: send boot/shutdown on first open connection, not blindly first in list
Replace break-after-first-iteration with next(c for c in connections if
c.transport) so the message goes to the first connection that actually
has an open transport. Falls back to connections[0] if none are open
yet (sendto will attempt reopen), avoiding silent message loss when the
leading connection is still connecting.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-04 09:59:30 -04:00
andreas c93dbdc0f4 fix: settings thresholds show correct per-config metrics; misc hbc fixes
Settings page: pass threshold_checker to http.start so the Threshold
Configurations section has data. Use threshold_checker's already-parsed
ThresholdConfig objects instead of re-parsing the raw nested YAML.
Named (non-default) configs now display only their explicit overrides
via threshold_raw_configs, not the full merged set with defaults.

hbc/hbc_mini: send boot and shutdown messages on first connection only
to avoid duplicate packets when multiple servers are configured.
Replace print("Daemonizing...") with logging.info so output goes to
syslog in daemon mode.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-04 09:12:39 -04:00
andreas 11d1c718b3 feat: retry AsyncConnection.open() indefinitely; drop IPv6 only on early startup failure
IPv4 connections are retried forever in heartbeat_sender if open() fails,
so a temporary network outage does not terminate the sender.

IPv6 connections that have never opened successfully are dropped after
IPV6_EARLY_FAIL_LIMIT (3) consecutive failures so that a network without
IPv6 support does not keep a dead sender running.

At startup all resolved connections are added to the list regardless of
whether the initial open() succeeds; the heartbeat_sender loop handles
the first real connection attempt.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-05-04 12:29:35 +02:00
Andreas Wrede 2bd3a9beb6 feat: restart on SIGHUP in hbc and hbc_mini
Sets dorestart and triggers a clean shutdown; os.execv re-execs
the process with the original arguments after cleanup.

Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
2026-05-02 10:06:26 -04:00
Andreas Wrede 462a445235 feat: add hbc_mini single-file client; drop dead connections on protocol error
- scripts/hbc_mini.py: self-contained hbc with no external deps; uses
  /proc for CPU/memory/network on Linux, df for disk, JSON config
- hbc + hbc_mini: mark connection _dead and stop sending on protocol error
- README: document hbc_mini usage, config, and plugin availability
- pyproject.toml: include hbc_mini.py in script-files

Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
2026-04-30 17:50:19 -04:00
Andreas Wrede c5ce41762e feat: update hbc via hb_install.sh instead of code patching
Server now sends a bare UPD command; client runs hb_install.sh to
reinstall from the package registry, then restarts. hb_install.sh
also copies itself alongside hbc on client installs.

Co-Authored-By: Claude Sonnet 4.6 (1M context) <noreply@anthropic.com>
2026-04-30 13:55:15 -04:00
andreas 65ceb31d8d fix: use os.path.exists check for /dev/log instead of dead-code OSError catch 2026-04-25 16:36:00 +02:00
andreas 1c9b6c1ca9 fix: reconfigure logging to syslog after daemonize() instead of no-op basicConfig
After daemonize() redirects stderr to /dev/null, the existing StreamHandler
writes to /dev/null. logging.basicConfig() is a no-op when handlers are
already configured, so log messages are silently lost.

Replace the daemon block to:
1. Call daemonize() first
2. Explicitly remove existing handlers (pointing to /dev/null)
3. Add SysLogHandler pointing to /dev/log with fallback to UDP localhost:514
4. Log startup message to the new syslog handler

Removes redundant syslog.openlog() call which is no longer needed.

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
2026-04-25 16:29:54 +02:00
Andreas Wrede ee3b72878f Add a ping monitor 2026-04-11 15:25:23 -04:00
Andreas Wrede 97dfc08f4d fix log level settiung 2026-04-10 08:00:51 -04:00
Andreas Wrede c5770006f7 hbc proper termination, hbd config reloadable 2026-04-02 07:17:00 -04:00
Andreas Wrede 460d2be9e9 Fix rtt, including bug in time compute 2026-04-01 19:41:53 -04:00
Andreas Wrede 079e84f729 display tag fro alterts, cleanup udp 2026-04-01 11:49:55 -04:00
Andreas Wrede 0543266c92 Major refactoring of the codebase, including restructuring of files and directories, renaming of modules and classes, and improvements to the overall organization and readability of the code. This refactoring aims to enhance maintainability, scalability, and clarity of the codebase while preserving existing functionality. The changes include:
- Restructuring of the project directory into client and server components
- Renaming of modules and classes to better reflect their purpose and functionality
- Moving common utilities and configurations to a shared location
- Updating import statements to reflect the new structure
- Adding new documentation files for better clarity on various aspects of the project
- Removing deprecated or unused code to streamline the codebase
- Ensuring that all existing functionality is preserved and that the codebase remains functional after the refactoring.
2026-03-29 11:13:40 -04:00