I have been playing with Pi Hole that is a server that reroutes add’s.
One of the (main) reasons why I installed it, is it gives me a view of what are all the different machines/programs in that are calling home (or other places)
To my surprise I saw that my openhab server did +300 calls to openhab.jfrog.io
and that in less then a few hours.
Maybe it has a reason why this happens. I can’t see it.
Will someone explain me why this is needed?
( I have openHAB 2.3.0 Build #1260 installed on a Pine64)
why would an installed and configured server need to contact the place where add-ons are hosted?
(we are at 495 now, its the most contacted domain from my whole network… )
I assume the online installs are still done when there are changes done to the configuration.
I did no change on the configuration last night.
While I was asleep, we now have 1140 connections…
I suspect that maven is configured to poll jfrog for changes since usually when it sees changes it can initiate a build to update your install. But OH doesn’t work like that. There won’t be any changes to that repo so the polling isn’t doing anything as far as I can tell. But I’m just guessing here.
@kai, is this expected behavior or something that is miss configured? If miss configured, is the openhab repo the correct place to create an issue?
From what I can tell from here all we would need to do is add an org.ops4j.pax.url.mvn.globalUpdatePolicy option to the mvn config or adding an update option to the jfrog URL in that config file:
Of course, this presumes that the default behavior is not desired. This is all far outside my area of expertise.
As for appbeat.io, @halloween, I suspect you have an add-on or some other program that is polling address. I cannot find that domain anywhere in my installs.
With netstat -tp you can see the processes with open connections. If you add -c it will update regularly.
Unfortunately it won’t help to identify a binding because it the connection is made by an addin the process name will be “java”.
I have searched on githup the different openHAB repos and did not find any reference to appbeat. Did you install any add-ons from Marketplace? Maybe somebody added calls to appbeat to collect information about the add-on usage.
Correct, openhab.jfrog.org hosts our snapshot online repository, while bintray.com hosts the one for releases (so if you use 2.2 you will see connections to bintray instead).
Note that you can configure whether the online repo should be used or not - this is done in the addons.cfg file (or alternatively in the Paper UI Configuration->System->Add-on Management). If this is disabled, I’d assume that no outgoing calls are made.
That has nothing to do with apt. The online repo simply contains all add-ons of openHAB, see http://www.openhab.org/downloads.html: “This is a zipped version of openHAB runtime. It is configured to install any add-on from a remote repository.”
I think the confusion is if these assumptions are true:
once I install for example OH 2.x SNAPSHOT xxxx OH will only install add-ons of the same version and build number
this means the add-ons are not being updated or changed for that specific version
the only time OH should need to contact jfrog is when installing a new add-on
So the question/concern is why OH contacts jfrog over a hundred times per hour (at least for yves)? It isn’t learning anything new and nothing is changing. That ends up being a whole lot of hits to jfrog doing nothing.
Or is there something I’m missing or don’t know about that makes Maven or whatever is doing the polling to check so frequently?
@yves, can you change the jfrog url in userdata/etc/org.ops4j.pax.url.mvn.cfg to match what I posted in post 9 above (with the @update=never appended to the end) and see if;