LibreHealth EHR Distribution

release
infra

(Art Eaton) #22

I can’t think 20TB big. 5000 downloads a month of our largest package amounts to 500GB. Even if we used up 5% of the total, I think we should be lucky to need more bandwidth.


(Art Eaton) #23

@sunbiz …and it is one of [INCLUSIVE OUR] VM’s. This is the beginning of the slightly broader LibreHealth.


(Saptarshi Purkayastha) #24

I suggest using hosted services for distributing things like packages for downloads. I would recommend that we use something like bintray, which usevCDNs and are much faster and robust to distribute downloads. Please join here - https://bintray.com/librehealth . for demos I think we can use the VMs


(Art Eaton) #25

I joined up, thanks for pointing me there. I take it this is a repo (among other things) that will let you serve out large file sizes…but I think I will need to do some homework before I do anything there. Glad to have downloads served out elsewhere, and I really think that doing an Apache Friends Bitnami install might be cool too…and an Ubuntu package or whatever.


(Robby O'Connor) #26

I recommend you use the server for a demo/your website. Not for distributing things.


(Robby O'Connor) #27

I also tossed Docker on there in case you want it.


(Art Eaton) #28

The only Docker I am familiar with make my shoes. Someone might know how to use it though. I ain’t actually a webdev.


(Terry Hill) #29

I thought they wuz pants


(Art Eaton) #30

They make those too…and shorts…well…someone somewhere makes them and then they stick their label on them. You know, like a plagiarist…but not so dishonorable.


(Robby O'Connor) #31

It’s freaking amazing:


(Art Eaton) #32

I gotta agree. I’m kinda partial to the cargo shorts with the 9" inseam myself, and they do look good with the shoes. Haven’t found any at goodwill for a few years though. Seriously, I am such a small-time operator that I doubt I will ever have any experience where I need to deploy large arrays of instances of anything. The deployment tool would be larger than the production instance! I’m just a…what…uh…duffer? Tinkerer? -Oh yeah! I’m a boatwright…not a professional webdev…so I am nigh-on useless. I only know code because it’s a tool in the engineering/fabrication bag. I only know medical practice management because other people needed someone to know it. I am pretty sure that Terry and I will have to start learning how the big-boy tools work, even if we are not deploying them ourselves. Don’t expect too much out of a couple of COBOL nerds too quick.:nerd:


(Robby O'Connor) #33

@aethelwulffe, I’m willing to help as much as possible!


(Art Eaton) #34

Once the Perl nerd…Tony gets settled we will probably have some meaningful input on what we want help with. Honestly, if we are only deploying a demo, and we have the code for you, we should be able to provide a domain name, the VHOST permission tweaks for the application, then let the spin doctors like yourself spin it in the right direction. Our master branch is ‘it’ at the moment…not even sure if we are going to do a REF:RELEASE / MASTER:DEVTIP or some other scheme…or switch to GITLAB… That still leaves the distribution side of things up in the air a bit. I didn’t catch on that this was mostly referencing the demo vs distribution until today. I will continue digging into BinTray.


(Tony McCormick) #35

I am anxious for us to get a handle on how we can use Containers (and Docker, Kubernetes) and other new tech to make our lives easier and our customer interactions and installs smooth…


(Terry Hill) #36

@aethelwulffe looks like a “Wardrobe change”.


(Robby O'Connor) #37

Have at it - @tony, are you able to get in? See the message between @aethelwulffe and myself


(Saptarshi Purkayastha) #38

I think we need to figure out using the gitlab CI’s new tools that allow creating docker images. They also have a nice container registry that is from the pipeline. I need to experiment a little more with lh-toolkit… its testing, building and deploying at the moment, but then moving on to distributing is the next step… The best practices can then shared by multiple LibreHealth projects.


(Art Eaton) #39

I think that not only do we need at least two people that are very familiar with each application we are using for project throughput, we also need discovery/tracking notes from the people learning as they go. This probably should apply to a wide range of things, not including helping to search for good used pairs of Docker shorts. This extends outside the realm of this thread, but I would like to see us pay attention to all levels of the full I/O product cycle. The code development part is being managed in a very appropriate way. Other parts of the process like market evaluation, and obviously things like deployment are a little out of scope right now, and traditionally have been by FOSS products. The really successful ones have policies like that. I doubt any of the unsuccessful ones do, almost to a 1:1 relationship. Having a policy that maps the whole process of serving the customer really helps pull you out of the tunnel vision that says “Code, Code, Code”. Everyone here knows that a little holistic awareness helps everything go smoother.


(Art Eaton) #40

I am really starting to feel like the EHR needs to move to gitlab…sooner than later.


(Robby O'Connor) #41

One thing…I’d like everything to be done in ansible…so let us know what you need. This is so we can easily move servers if needed

Deploy using CI :slight_smile: I’ll add a CI user to all servers