So let's say you have something you want to run in a console application, perhaps something with a few jobs for an application. So simple answer is, write some code in the Program.cs and away you go... Now, let's say you need to do three different things and for performance reasons they can all be done simultaneously and when they all are completed, you want to execute something to know they are all done. There are a lot of ways to do this with DI or with different toolkits, but without the full pop and circumstance, here is a simple method to do this. The only using
Now you get the exception "The configured user limit (128) on the number of inotify instances has been reached" and you wonder what the heck is going on. For me I never got this exception on my local instance of Docker running on my Mac or in Windows but I would see my app just stop running in Linux. Even weirder I could run the app fine in Windows hosted by Kestrel. So where does this error start coming from?
After much digging, I found the problem was with the IConfiguration interface. Specificially, I was calling pulling the JSON configuration files multiple times which was reloading the IConfiguration interface. The issue was that it was rebuilding the configuration and re-attaching the "reloadOnChange" setting. Since the reloadOnChange was set to true it was attaching a new file watch multiple times and was hitting the limit. That all being said, I never saw this issue locally on my Mac or Windows or my local instances of Docker on either which I think is related to the way the underlying Docker Linux host manages the watch on the files.
Something else as a side note, since the files were being tracked, even though I was disposing of the class it wasn't releasing the watch and thus keeping these connections open. The simple fix was to set the reloadOnChange to false but to make it even better was to just create a singleton of the data and thus prevent the data from being ever reloaded. I don't think it is necessarily a bug with .Net but it is a weird side effect of how the host can cause problems with a framework.
I hope this helps some people save some time that I lost dealing with this crazy issue.
I like many developers am tasked with creating batch emails to send to a large population of users. And for this I am not talking about spam, I am referring to sending emails to a large poplulation within an organization such as annual notification reminders or custom links to some application. I have tried sending these emails synchroniously and if you are sending a few hundred that isn't a problem, but when I started sending several thousand, I was noticing a bottleneck with my SMTP server and waiting for the callback showing the SMTP server receved the message.
So moving forward from that thought the best option for sending mass emails greater than a few hundred is using threading and asynchronious sending. I have tried this a few different ways in the past with event listeners and different techniques, but moving to the async and await world of modern .Net I thought I would re-write it with some more modern tooling. Since this took me a little trial and error I thought it would be a good thing to post for my fellow developers out there.
From a performance standpoint using the same SMTP server I was getting about three emails per second with standard synchronous methods and when I went to the attached code, I went up to about 18 (running locally over VPN to our corporate offices) and I will expect to see an even bigger jump on a VM in the datacenter. Overall there isn't much impact to performance on the machine with minimal CPU and RAM overhead.
Anyway, click here to get to my code.
Two things I want to point out:
- I know I have a weird try / catch in there. That is because the mail server was choking with that many emails going through at one time and I was getting random errors. Retrying one time usually fixed it.
- I know there are async methods with the SMTP client, but when you use that it still only allows one email to be processed at a time by the general SMTP client thus making it an async method but not async in the since that I wanted which was mass emails with threading.
Anyone familiar with Yammer will know that many of the APIs are a bit lacking. Also, all API connections will need an oAuth token that has access to the data in question. So what do you do if you have a console application that needs to use Yammer Data Export API and drop the into a database or more to the point, how do you get the oAuth token? I am not going to post the specifics on how to get the data and extract it but I will post some code on how to get the oAuth token.
To start there is some information you will need from Yammer which is only obtained after creating a Yammer app (https://developer.yammer.com/v1.0/docs/app-registration). Your redirect URL for the included code needs to be http://localhost:80/Temporary_Listen_Addresses/ (if you get an error relating to access denied the port may be being used or it could be an issue with your privileges on the device). *NOTE you do not need to publish your app for this functionality to work, so you can create the app then never finalize it. Yammer's documentation https://developer.yammer.com/v1.0/docs/authentication-1 does not include the specifics on how to implement this on a console application and only includes the basic information.
The included code makse the assumptions that you will have the following app seetings in your App.config: "ClientID", "RedirectURL", "ClientSecret", and "oAuthToken" with all of these fields populated, with the exception of the "oAuthToken" field, from the information in your Yammer app settings.
I have included the complete class to perform the login here and to implement it you just need to call the constructor.
- string token = yammerLogin.oAuthToken;
The oAuthToken will be stored in the App.config and it will only re-auth if the token has failed to login.
I hope this will help save a few hours of work for a few people out there.