driver / standard (pull_request) Successful in 15s
driver / lint (pull_request) Successful in 13s
driver / types (pull_request) Failing after 56s
driver / test (3.12) (pull_request) Failing after 46s
driver / lowest (pull_request) Failing after 20s
driver / test (3.13) (pull_request) Failing after 47s
driver / build (pull_request) Successful in 27s
driver / security (pull_request) Failing after 1m4s
CI / driver (pull_request) Failing after 2m18s
driver / publish (pull_request) Skipped
napalm-linux now brings napalm-device-types' POSIX scheduled jobs, with Debian's paths. On QTS they are wrong: the crontab that survives a boot is /etc/config/crontab (admin's, no user field), crond runs a copy of it, and a change made only through crontab is gone after the next boot. BusyBox crond reads no /etc/crontab, cron.d, run-parts or anacron, and there is no systemd. QtsScheduledJobsMixin reads /etc/config/crontab as admin's and the other users' crontabs from crond's spool in one round trip, and writes the file first, then loads it and restarts crond. The firmware's own lines are platform by a conservative command list and only run, until a real box shows whether QTS leaves one switched off alone. An application's job is marked lines in admin's crontab; a run is bounded without timeout, which QTS's BusyBox cannot be relied on to have. Requires napalm-device-types 4.26.0 and napalm-linux 0.16.0. Closes #17
5.4 KiB
5.4 KiB
Changelog
All notable changes to this project are documented here.
The format is based on Keep a Changelog, and this project adheres to Semantic Versioning.
[Unreleased]
Added
- Scheduled jobs on QTS (#17, NetOrk/netork#1933). napalm-linux now brings
napalm-device-types' POSIX scheduled jobs with Debian's paths, which are wrong on QTS;
QtsScheduledJobsMixinputs them on QTS's:get_scheduled_jobs()reads/etc/config/crontab-- the crontab QTS keeps across a boot, admin's (uid 0), without a user field -- and the other users' crontabs in crond's spool, in one round trip. crond's copy of admin's is not read twice. No/etc/crontab, cron.d, run-parts, anacron or systemd: BusyBox crond reads none of them. The time zone is QTS's own (getcfg System 'Time Zone') when it is an IANA name. A box without/etc/config/crontabraises rather than reporting no jobs.- The firmware's own lines (
/sbin/hwclock -s,/sbin/qfstrim,/etc/init.d/backup_conf.sh, ...) areplatform, by a conservative list of the commands stock QTS writes; a line that also runs something else stays the admin's. They only run: QTS's scripts and UI keep these lines, and whether they leave one switched off alone needs a real box (#13). manage_scheduled_job()switches the admin's lines and runs any line now. A write goes into/etc/config/crontab(cp, so the file keeps its mode, owner and links), thencrontab /etc/config/crontaband/etc/init.d/crond.sh restart-- without them QTS forgets it on the next boot. A run is bounded bysleepandkill, nottimeout, which QTS's BusyBox cannot be relied on to have.put_scheduled_job()anddelete_scheduled_job()keep an application's job as marked lines in/etc/config/crontab, never in a cron.d. Only as admin (rootis taken for it): every other crontab lives in RAM and is gone after a boot.
- Requires napalm-device-types 4.26.0 and napalm-linux 0.16.0.
Changed
- Personal accounts are switched off (
get_local_accounts,add_local_account,lock_local_account=None). napalm-linux now adds and locks accounts withuseradd/usermod(NetOrk/netork#1793); QTS manages its users in its own UI and ships no shadow-utils. A QTS variant follows once it can be built against a real box. - netOrk's own account (
add_service_account, NetOrk/netork#1817) is switched off for the same reason. set_hostnameis switched off (= None). napalm-linux now renames a host throughhostnamectland/etc/hostname(NAPALM/napalm-linux#45), which QTS would report as done and lose on the next boot: its name lives inuLinux.conf,/etcin RAM. netOrk then says the name is kept in netOrk only. A QTS rename follows once it can be built against a real box (#13).- Python 3.12 is the floor (
requires-python = ">=3.12"): netOrk runs every driver on 3.12, so 3.10 and 3.11 are no longer tested (NAPALM/driver-standard#6).
[0.6.0] - 2026-10-23
The first release on PyPI.
Changed
- The driver follows the NAPALM driver standard
(https://git.netork.io/NAPALM/driver-standard): CI on Gitea runs ruff, mypy, the
tests on Python 3.10–3.13 and at the lowest declared versions, bandit, pip-audit and
a build check, and a release tag publishes to PyPI. The package now ships
py.typedand a LICENSE file. It needs napalm-device-types 4.22.0 and napalm-linux 0.15.0. Python 3.9 is no longer supported; napalm-device-types never ran on it.
[0.4.0] – 2026-10-17
Changed
- Requires napalm-linux 0.8.0, whose fixes are napalm-device-types' templates.
configure_snmp_agent,configure_package_proxyandensure_packageare switched off (None): QTS has no apt, dnf or opkg, and its SNMP agent is set in its own UI. Before, the inherited SNMP and apt-proxy fixes were offered and failed.
Fixed
get_device_warningsno longer logs a TypeError on every poll: the inherited one asked for updates, which QTS cannot list (#6).
[0.3.0] – 2026-10-08
Removed
- The
_docker_bin()override: napalm-linux 0.3.0 drops the Docker CLI methods it served._container_engine_binary("docker")returns Container Station's path itself.
Changed
- Requires napalm-linux >= 0.3.0.
[0.2.0] – 2026-10-07
Added
_container_engine_binaryreturns Container Station's docker path, so the container engine connection from napalm-device-types 2.6.0 (open_container_engine("docker")) reaches it although it is not on PATH. Callers pass arguments only and never see the path (NetOrk/netork#765).
Changed
- Requires napalm-device-types >= 2.6.0 and napalm-linux >= 0.2.0.
[0.1.0]
Added
- Initial driver scaffold:
QnapQtsDriver(StorageDriver, LinuxDriver)with discovery fingerprints (SNMP enterprise OID 1.3.6.1.4.1.24681, HTTP, port specs) and theqnap_qtsentry point. - Explicit
DEVICE_CLASS = "storage"so netOrk does not classify a QNAP as a plain Linux host and hide its Storage tab. - QTS major-version detection, Container Station docker path discovery and
Virtualization Station virsh discovery, all resolved once per session in
open(). - MRO collision resolution for
get_services, whichStorageDriverwould otherwise shadow. tools/harvest.shandtools/sanitize.pyfor capturing and anonymising command output from real hardware.
Pending
- Storage, QPKG and VM parsers — blocked on the fixture harvest.