2023-07 Time marches on and I think I can make this work as a node server, then I can use (easily) graphql. Or I can make it a flask app still using graphql.
2022-06 I am working on turning this into a microservice instead of a full web page, so the web page you can load at port 5000 is no longer as pretty as it used to be. (See the notes on version 2 all the way at the bottom of this page.) I got arctic-monitor to a useful stage and now I work on it when time permits.
System for monitoring an ArcGIS concurrent license manager
This is a small Dockerized app written in Python running as a service. The Python script needs to query the license manager and it does that using the "lmutil" command line utility that is included in the Esri download for the license manager.
Because the license manager "Flexera FlexLM" is licensed software, if you want to use this monitor in the Docker, you will need to download the Linux version of the license manager package from ESRI and put it here in the project folder before doing the Docker build.
The Dockerfile is based on this version of the license manager: ArcGIS_License_Manager_Linux_2022_1_184756.tar.gz
I don't think it matters very much which version you use, because it is just interrogating the real license server over a network connection. But when the version number changes you can change the Dockerfile.
To get the tar file, go to my.esri.com and download the latest Linux license manager. It will be a file ending in 'tar.gz'. Put the file in this folder. (The one containing the Dockerfiles.)
The requirements doc at ESRI call for RHEL 6 or 7; this Dockerfile uses Centos 7 and Debian 11.
I took the list of required RPM packages from the ESRI documentation and dropped them into the Dockerfile as "stage1".
I broke up the package installation into multiple RUN commands. That's because when I tried doing them in one pass, it failed. I did not try to fine tune it or confirm which ones were really needed or what order to install them in once I got it working.
The license manager installation step is done in "silent" mode so there is no requirement for any X Window server or any interactions from you.
I don't like Redhat, so stage 2 of the Dockerfile uses Debian 11. The final image is smaller. I am sure it could go a lot smaller but I am done for today. Maybe someone could make it run in Alpine?
Because of the licensing constraints I don't push any image file up to Docker Hub.
Make sure you've downloaded the tar.gz file, see Prerequisites.
Then run the build command to create images for the license manager and the monitor.
Install service.txt -- In the interest of simplicity you have to put a copy of your service.txt file from your license server here so that it can be baked into the build. You need to edit the service.txt file so that it has the actual license server host name instead of "This_Host". Copy the service.txt file into the config/ directory, and edit it.
Build the image. Use buildx or don't. I find it far faster than using docker-compose
docker buildx build -t wildsong/arctic-monitor .If the build fails with a message about not being able to ADD then you did not put the tar.gz file here or you need to update its name in "Dockerfile" around line 55.
After the license manager is installed Docker will emit a long series of Copy File and Install File messages from the flexlm installer. It will stop at this point if the install fails.
For the monitor, only file we need from the ESRI installation is lmutil. When the stage 2 image is built, the file will be copied from the stage1 image to the stage 2.
Once the builds complete you will have an image called "wildsong/arctic-monitor" with the lmutil tool and the python web server.
You can look around in the new container by launching into a bash shell. If you don't want to, skip to the next section.
docker run -it --rm wildsong/arctic-monitor bashWhen you are in the shell you can run "./test.sh" and it should dump out the license info. You can run "python read_lmutil.py" to make sure it can run the Python as well.
You just have to run the container. Docker-compose.yml has the environment set up and is set to restart so use that.
docker-compose up -dI previously started working on a Windows-based monitor and quit when I found out how hard it was (FOR ME) to work with Docker On Windows.
Uses lmutil and stores output in SQL Server: https://github.com/jmitz/ArcGISLicenseMonitor/blob/master/LicenseMonitor.py
14:46:39 (telelogic) DENIED: DOORS indkach@indkach [telelogic] (Licensed number of users already reached. (-4,342:10054 )) 14:46:39 (telelogic) DENIED: DOORS indkach@indkach [telelogic] (Licensed number of users already reached. (-4,342:10054 )) 14:46:39 (telelogic) OUT: TLSTOK-token indkach@indkach [DOORS] (3 licenses)
REPORT LOGGING - can be enabled in OPTIONS file but produces an encrypted file that is of no use without Flexera software
The advantage of doing this is that I can see the log file and respond as soon as anything happens instead of polling. The disadvantage is that if the server is down, the monitoring web site will be down too. Further it precludes multiple redundant license servers.
If I put a microservice on the actual license manager, and let it run both lmstat and monitor the logs, and then make a dumber flask app that does not need to have lmstat installed, I will be several steps ahead. So I will be doing that on the "dev" branch of this project starting right this minute.
The microservice would be a MQTT publisher instead of running over HTTP, the pub/sub model makes more sense here.
Run a process via exec https://www.digitalocean.com/community/tutorials/how-to-launch-child-processes-in-node-js