Here is the network tab: https://imgur.com/5iEX0Qa
Proxy settings: https://imgur.com/4vtoQOr
its weird if i refresh it works, if i refresh again it breaks again
Hi there, I have a question about using geohealthcheck with docker, I'm not sure if it's an issue or if my understanding is at fault. My problem is that if I pass a SCRIPT_NAME environment variable with the docker run command, whatever path I include seems to get used twice. My run command is:
docker run -d --name ghc_web -p 8083:80 -e SCRIPT_NAME=foo -v ghc_sqlitedb:/GeoHealthCheck/DB geopython/geohealthcheck:latest
... but then when I load the page at http://localhost:8083/foo then all the references in the page are to http://localhost:8083/foo/foo/?lang=en
If I don't include a SCRIPT_NAME variable then it loads as expected at http://localhost:8083, but unfortunately I need it to run in a sub directory
is this a bug or have I missed something in my understanding of how this should be working?
Well, first of all
SCRIPT_NAME is meant for the case where GHC is run behind a proxy like Apache or Traefik which you probably intend, though it should work in isolation as well. I remember this question came up before, and you contributed a nice piece of documentation for that case: https://docs.geohealthcheck.org/en/latest/install.html#running-under-a-sub-path . Is it maybe you forgot a slash i.e. /foo i.s.o. foo?
I can't get the nginx config to work (using my documentation from earlier :-) ) because it is creating the nested sub-directory. I have an older version (0.6 ish) working outside of docker with gunicorn and supervisord and that is working just fine and the nginx config ought to be the same
but... surely it should work in isolation (ignoring nginx) so adding a script name as an environment variable should not create the nested sub directory- it should just create one?
If I set -e SCRIPT_NAME=foo then the URLs are of the form http://localhost:8083/foo/foo/?lang=en when they should presumably be of the form http://localhost:8083/foo/?lang=en
ah- so docker run -d --name ghc_web -p 8083:80 -e SCRIPT_NAME=/foo -v ghc_sqlitedb:/GeoHealthCheck/DB geopython/geohealthcheck:latest seems to work
I'll try that in my server environment
Good day, I recently install GeoHealthCheck and I am in the process of adding resources. I have an error when I try to add probes to a WFS resource. For example, the probe WFS GetFeature in BBOX for ALL FeatureTypes give me an ‘Error getting Probe form for GeoHealthCheck.plugins.probe.wfs.WfsGetFeatureBboxAll’. The log error message says: ‘UnboundLocalError: local variable 'feature_type_entry' referenced before assignment’ for ‘crs_list = feature_type_entry.crsOptions’. I am using 0.8.dev0. Someone has an idea why I have this message? Thanks a lot!
Ok, I cannot reproduce on de demo.geohealthcheck.org site (runs latest
master) using a public WFS from the Dutch SDI: http://geodata.nationaalgeoregister.nl/bag/wfs . But I can see where it happens in the code. There a random feature type entry is fetched but it is appearantly empty (code should be more defensive though). I see two possible causes: 1) your WFS provides different options/encodings as e.g. the one above. Or 2) a mismatch with the OWSLib version you are using. So can you give the WFS URL you are using? And the OWSLib version?
I use OWSLib version 0.17.1. The WFS I use is password protected, there are no layers that can be accessed without being logged in. Is it possible that crs_list is done on an empty WFS (before authentication is performed)? I tried the public WFS from the Dutch SDI and it worked on my instance.
OWSLib 0.17.1 is ok (same as used on demo). The auth should not be the problem. All info is pared from the Capabilities doc, guess the Capabilities Probe works ok. As the public WFS URL works, there is something in the capabilities doc different. For one thing: I see that the WFS Probe fetches using WFS 1.1.0, could that be a problem (e.g. only 2.0.0 is supported). You may want to look into the Caps doc, it should read like:
The general structure of the GetCapabilities document is the same. However, I see a difference in the CRS. There are in my file 2 x ':' between EPSG and the code when there's only one in the public WFS from the Dutch SDI Caps. For example, '<DefaultCRS>urn:ogc:def:crs:EPSG::4326</DefaultCRS>'. I use GeoServer to publish data.
That looks correct. Another public WFS uses same notation (double colon): https://geodata.nationaalgeoregister.nl/nationaleparken/wfs with
<DefaultCRS>urn:ogc:def:crs:EPSG::28992</DefaultCRS> and works ok. Give that a try. Must be something else. Hard to figure out without the WFS url and your deployment context (Docker?, Python version, maybe lxml version). The capabilities Probe and other Resource types, w.g. WMS work ok for that GeoServer endpoint?
Hmm and https://geodata.nationaalgeoregister.nl/bag/wfs now suddenly also has double colons in CRSs.
But that is the correct notation.
See also: https://docs.geoserver.org/latest/en/user/services/wfs/axis_order.html
urn:x-ogc:def:crs:EPSG:4326 and urn:ogc:def:crs:EPSG::4326 are equivalent.
(notice the subtle
:x-ogc: vs :ogc:, 4 notations for the same thing causes headaches in many projects).
Forget the above, the problem (and fix) is clear to me, and already known: https://github.com/geopython/GeoHealthCheck/issues/9#issuecomment-562950148 :
WfsGetFeatureBboxAll uses OWSLib WFS, which in 0.17.1 does not support auth headers. Was fixed geopython/OWSLib#613, even by me :-), in later version. There is even a geopython/GeoHealthCheck#312. (WFS Capabilities Probe works as it does not use OWSLib btw).
An activity for a GHC API specification has been started. In first instance geared to facilitate remote monitoring integrations and custom reporting, using the current GHC implementation. There is a broader plan to base a future GHC entirely on a full CRUD API. See work-in-progress on the Wiki: https://github.com/geopython/GeoHealthCheck/wiki/API-Specification . If you would like to comment or have suggestions, please do, best is via the related issue #325.
Would like to bring out GHC 0.8.0 today. This is mainly the Py2->Py3 migration, no DB upgrades. Has been running stable on "demo" for about 6 months now. Several production deployments can then be upgraded (on 0.7.0). Then we can quickly move to Milestone/version 0.8.1 that has leftovers from Milestone 0.8.0. @tomkralidis I can do the release sequence ok?
@/all GHC release 0.8.0 is out. Main addition is the Py2->Py3 migration. Many thanks to @borrob for his efforts/patience with this! Also 0.8.0 Docker Image is available.