FYI. Just updated MeshCentral Router for Windows with a new relay port map mode. You can now use a remote agent as a TCP relay to access other services on that network. Feedback on it appreciated.

You guys get an premiere, I am posting this here first. I just published the all new MeshCentral Router for Windows. Allows TCP port mapping thru the MeshCentral server. Requires v0.3.0-t or later. You can RDP, PuTTY, WinSCP to remote computers. Youtube video demonstration here. Feedback & bug reports appreciated.

@pmoncho Arg. I need to investigate this NPM issue. I have not seen this problem on my own servers, but it may be an issue with a specific NPM version. I will take a look.

@bnrstnr When you update the MeshCentral server, it will auto-update all the agents when needed.

We got cut & paste almost done on both Linux and Windows. Most of the code is already these, just turned off until we gets it working well. Hopefully in the next week.

Everything back to normal now. Sorry about that. Server heap memory optimizations gone wrong. There does seems to be a slow memory leak on the server and I am tracking that down.

v0.2.9-z is out. That should solve the fires. Desktop/Terminal/Files working again.

There is another problem, going to patch ASAP. The remote desktop/terminal/files does not work.

Published v0.2.9-y with a fix for this.

Good input. Working on a few big ticket items right now (Agent update recovery and Intel AMT CCM activation), but user notification & consent is a frequently asked feature. Will try to give it a first go at it next week.

By the way, I love Red Dwarf I am terrible at SMEG... I need to do more SMEG'ing for my own sanity. Pushed out more MongoDB optimizations.

@JaredBusch You guys are demanding! Good feedback, entire checkbox label is active in v0.2.8-x I just published.

@JaredBusch I just published v0.2.8-v with support for batch move of devices between groups. I also did quite a bit more work on database optimization. I will check out the rename/screen things when I get a chance. Not related, but try holding shift when you press the full screen button.

One more quick news. I have been working on MongoDB tuning in the last few days. You should be able to connect 10k+ devices easy on a AWS t3.medium instance (Amazon Linux 2, 30$/mo + traffic) without issues. Just published v0.2.8-r with proper MongoDB indexes and query fixes. Handles 1000's of devices now. Still room for improvement, but the previous version where REALLY bad with connection scaling. v0.2.8-r is a lot better at that. Graph below shows MongoDB wasted time for a server with 5000+ agents.

@JaredBusch Thanks. Yes, you can now change groups. By the way, maybe this will help your installer: You can create a text file with the same name as the agent with ".tag" at the end (or instead of .exe on Windows) and put some serial number or short unique id in it. When the device shows up, click on it and your will see "Agent Tag" with the content of the tag file. This is useful if you want to put your own device mapping ID and may help you sort devices.

@StuartJordan On the topic of the evil "/createLoginToken.ashx?user=(username)&pass=(password)". Good feedback. I do want to allows people to setup MeshCentral on a Raspberry Pi, etc and tinker with it on a local LAN or to help developers get started. However, since I don't want anyone using this in production, in the next release: It will only be allowed if you are using the default un-trusted HTTPS cert. This way, if you start using a Let's Encrypt TLS cert, your own cert or a TLS offloader "createLoginToken" does not work anymore. Hopefully that will keep everyone happy

@vpr00 If you want to be completely evil... you can try on your server /createLoginToken.ashx?user=(username)&pass=(password) and it will return a login token. It's evil because passing credentials in a URL like this is really bad. Often URL's are logged and so your password is in event logs. This said, people kept wanting me to add this feature...

@black3dynamite Right, it's not always big updates. Doing a lot of security work recently. Just about to publish a new version with better password change dialogs, random time delay if on bad passwords, ask for old password when changing it, display password policy at the right places, etc.

I am running way behind on documentation. On embedding, let me know what server type will be the master (NodeJS, ASP.NET...). Once you get MeshCentral running, get the --loginTokenKey and cut and paste it into your server. Then, if the master server is NodeJS, use this code to generate a time limited cookie on your server and use it as documented. The user name is "user/(domain)/(account name in lower case)", the sample below is for "admin" on the default domain. hope it helps until I get time to work on documentation.

obj.crypto = require('crypto');
obj.encodeCookie({ u: 'user//admin', a: 3 }, obj.loginCookieEncryptionKey)

// Encode an object as a cookie using a key using AES-GCM. (key must be 32 bytes or more)
obj.encodeCookie = function (o, key) {
    try {
        if (key == null) { key = obj.serverKey; }
        o.time = Math.floor(Date.now() / 1000); // Add the cookie creation time
        const iv = Buffer.from(obj.crypto.randomBytes(12), 'binary'), cipher = obj.crypto.createCipheriv('aes-256-gcm', key.slice(0, 32), iv);
        const crypted = Buffer.concat([cipher.update(JSON.stringify(o), 'utf8'), cipher.final()]);
        return Buffer.concat([iv, cipher.getAuthTag(), crypted]).toString('base64').replace(/\+/g, '@').replace(/\//g, '$');
    } catch (e) { return null; }
};

Someone just posted the login problem on GitHub and I posted a fix for it. MeshCentral v0.2.7-r is now live. This fix should solve exactly the problem you got.