hassio-addons: 🐛 [Portainer] Unreliable since v2.17.1-4 - supervisor entities regularly unavailable
Which addon?
- Addon name : Portainer
- Addon version : 2.17.1-4
Describe the bug
Since running v2.17.1-4 (before I was using I think v2.17.1-2 - lost overview with this intransparent micro release updates), these supervisor provided entities are regularly unavailable:
sensor.portainer_cpu_percentsensor.portainer_memory_percent
This only happens for the Portainer addon, others are just fine.
To Reproduce
- Update to Portainer v2.17.1-4
- Do nothing for some time
- Watch listed entities switching to
unavailableregulary, turning back to normal operation after few minutes
2023-03-04 06:37:00.300 ERROR (MainThread) [homeassistant.components.hassio.handler] Timeout on /addons/db21ed7f_portainer/stats request
2023-03-04 06:37:00.302 WARNING (MainThread) [homeassistant.components.hassio] Could not fetch stats for db21ed7f_portainer:
-sensor.portainer_cpu_percent
-sensor.portainer_memory_percent
-sensor.portainer_cpu_percent
-sensor.portainer_memory_percent
Full addon log
Nothing in there corresponds to any time code the issue happens.
Full addon config
Default config, only password is set of course.
System
- Supervisor version: 2023.01.1
- Host system version: 9.4
About this issue
- Original URL
- State: closed
- Created a year ago
- Comments: 16 (8 by maintainers)
Commits related to this issue
- Rollback code to 2.17.1-2 https://github.com/alexbelgium/hassio-addons/issues/745 — committed to alexbelgium/hassio-addons by alexbelgium a year ago
- Update CHANGELOG.md https://github.com/alexbelgium/hassio-addons/issues/745#issuecomment-1460107531 — committed to alexbelgium/hassio-addons by alexbelgium a year ago
That’s a good theory. 3 days ago I updated the File Editor addon. Since that and today intensively I get alerts on that one… no idea what Supervisor messes up here. Currently I tend to say: probably not Portainer specific.
Thanks ; then we are aligned 😉
Re. versioning, HA only checks if the version name is different and not if it is incrementally increased. That’s why in your screenshot, the supervisor proposes you the
2.17.1-2as an “upgrade”. The2.1-7.14version should have been more reliable, given that it implemented the new s6-rc.d services system, plus periodic healthchecks. The-2version (on which future versions will be based) on the other hand did not contain the healthchecks, plus relied on the legacy folders (/etc/cont-init.d for oneshot scripts and /etc/services.d for services). The s6-rc.d system still calls on those folders through its legacy system to avoid code breakage slightly more detailed description hereWell, as long as there is no visible downsides I’ll keep it that way then. It is often very hard to discriminate issues stemming from addons from the ones originating from the HA os updates
It’s just a never ending story, happening many times a day.
Of course I could just put those sensors to the ignore list, but silencing actual issues never seems to be a good idea to me.
I found a backup from before updating Portainer and am gonna try to restore it and see if it changes.