If they can use VoIP, they can do hosted. I'm sure that hosted would be cheaper than the appliance, even over a really long period of time.

No Linux utility is more well known than the venerable top command. While many people have moved on from top to more modern tools such as htop and glances, top is still the main "go to" tool for instananeous system monitoring as it is light and widely available on essentially any Linux installation. All major distributions have it in their repos and essentially all install it by default. It is the one tool of this nature that, with relative confidence, you can be sure will be available to you at all times.

Using top shows a lot of information, here is an example standard output from a Fedora system:

As you can see, we have a lot of data at which to look, even by default. This display updates ever 1.5 seconds by default. nearly everything in top can be configured, but most people stick to just a few basic use cases. To learn more about top, just use the "man top" command in your distribution to read the extensive users manual.

The header section gives us a system summary including the system time, uptime and load factor averages. This is the same load information that we get from the uptime command.

We also see tasks, which shows us general process statistics including how many we have, and what states they are currently in. The total number of tasks should mirror the number if the ps aux command. This is the source of that data (or they share a source in /proc.)

The CPU line shows us the current CPU activity in a standard way.

The Memory and Swap lines show us basic memory statistics in real time, as well. This does not have data pulled out for us like free does, so this takes a little more effort to read. See the free page in this guide for a better understanding of how free handles this data.

The real meat of the top command is in the process list. This, by default, is sorted by the amount of CPU time used during the monitoring period. A quick look at this chart shows us:

PID: The Process ID of the process (task) in question.
USER: The system user that owns the process.
PR: The priority of the process.
NI: The nice value of the process.
VIRT: The virtual memory size of the process.
RES: The resident set size of the process.
** SHR**: The shared memory size of the process.
S: The process state with S for sleeping, R for running, Z for zombie.
%CPU: The percentage of the CPU used by the process (by default, this is the sort order of the list.)
%MEM: The percentage of all memory used by the process.
TIME+: The total accumulated processor time used by the process since being spawned.
COMMAND: The command that spawned the process initially.

Processes which are currently running (the S field is an R) are highlighted.

With this information, top tells us a lot about what is going on on our system. There are two really common options that make top even more powerful.

Sort by Memory Usage: Press "M" while top is running and the process list will immediately resort by the %MEM field instead of by the %CPU field. This can be really helpful in identifying processes hogging memory. Press "P" to return to the default view.

Show CPUs Individually: Press "1" while top is running to toggle between viewing all thread processors (labelled as CPU) aggregated in a single line in the header to having each logical thread processor broken out to show individual activity. This can be extremely helpful in identifying if a process is capped because it has exhausted a single thread capacity while the system as a whole may be mostly idle. When toggled to viewing each logical thread processor individually the %CPU field switches from displaying the number as a percentage of the whole to display the percentage of a single thread processor. So what might be using 5% of CPU in a eight LTP system, will show as 40% when this mode is toggled on.

Here is the stock view with all LTPs aggregated into one line:

And here it is with them broken out:

You exit the top command by hitting "q" to quit.

---

You can explore many additional options in top that you may find useful. It is an extensive monitoring system. For example, you can, with "m" toggle between different memory views:

Or you can use "T" to sort the chart by total CPU time.

---

Part of a series on Linux Systems Administration by Scott Alan Miller

Youtube Video

Too often, people jump to the conclusion that vendors are totally insane or have no clue what they are doing when, with a little thought and empathy, would often find that their actions do make sense and we've just not realized something or are not aware of all of the facts. Vendors get things wrong, certainly. But just assuming that big companies with a lot on the line do crazy things isn't a productive way to think about problems and will often lead to excuses and missing obvious answers.

@tim_g said in Fedora 26 No Space Left on Device with Plenty of Space Available:

Seems like a major Linux turn-off if this happens for no reason.

It's not Linux related. It's filesystem related. And it is so insanely rare that it's not a thing to worry about in the real world. Windows has inodes too, they call them File ID.

How big is the network? Sometimes just manually auditing into a spreadsheet is best.

Youtube Video

Understanding risk is one of the most important things that we do in IT. One of the best tools that I have found for describing the concept of "getting lucky" in risk is having your mother explain it to you. The "call your mom and tell her you don't wear a seatbelt and it's fine because you are still alive" problem does more to explain risk concepts than anything I've ever found. It's downright effective.

Youtube Video

Hyper-V has so many myths around it. Getting it installed correct, the first time, is important in building a reliable, long term virtualization infrastructure.

Having just released yesterday, it is time to get Microsoft's SQL Server 2017 installed on CentOS 7.4. I have a CentOS 7.4 VM fully updated and ready to go on my Scale .

First we add the MS SQL Server RHEL repository to our system.

curl -o /etc/yum.repos.d/mssql-server.repo https://packages.microsoft.com/config/rhel/7/mssql-server.repo

Then we just do our normal YUM based installation commands.

yum install -y mssql-server

We then configure using the build in command line configuration utility:

/opt/mssql/bin/mssql-conf setup

Once installed, MS SQL Server is managed via the standard systemctl commands.

And that is all folks. After that it is up and running. Worth noting, you will need 4GB of RAM to even fire up the process, which is a tad excessive when you consider that you can run its competition in production with far less RAM than that. And only XFS and EXT4 filesystems are supported. That means no ZFS, no BtrFS, no JFS2, etc.

Why are you talking to them? Move your DNS to CloudFlare. Done.

I had to add a video response as well. (I should mention that the video was uploaded via WiFi.)

Youtube Video

Zimbra, the leading open source email server, has finally released version 8.8 which includes, we are told, their new web interface. Zimbra 8.8.5 can be downloaded from their site right now.

ReadyNAS and Synology, two bay units. Diskless, add your own disks.

@nerdydad said in Need a host recommendation:

@scottalanmiller said in Need a host recommendation:

Why two hosts if you are forbidden to use them for HA?

He doesn't understand HA. He prefers IPOD because that is what he is familiar with.

That's like being familiar with a car crash so refusing to drive on the road.

FreePBX 14, found this issue today. A site that normally works will suddenly have all phones disconnect. I connected my own line from home to see if it would disconnect as well, but it registered and was solid even when the main site disconnected. All phones, all at once. I suspected a site network issue, of course.

Inbound and outbound trunks were always still working, so the issue had to be related to the site.

I found that if I watched the system long enough, that the phones would connect again, but not for long. All coming back on, or going off, together. I tried restarting the responsive firewall, and this would cause the phones to all connect, but then all disconnect again. This seems odd as the LAN has a static IP for that site and is whitelisted in the firewall, so in theory there should be no firewall activity to block the site.

I watched the logs from asterisk and noticed a lot of failed attempts at connecting from a device at that site with a bad password, which led me to suspect the intrusion detection system.

Sure enough, the issue was being caused by a fail2ban detector seeing the bad passwords and banning the site over and over again. Intrusion Detection in FreePBX is not under the Responsive Firewall, but is its own thing. To modify it, you go to Admin -> System Admin -> Intrusion Detection. Here you can see what IPs are currently banned, change any settings, set up and most importantly, whitelist any sites that you need to not have be affected by fail2ban's detection mechanism.

There has been much consternation around NextCloud being difficult to install, and scary for newcomers, especially as it is only available on Linux. Not only can it be difficult to install, but even just choosing which version of Linux can be a stumbling block if someone is not used to the decision process here. All is solved here with an enterprise grade, fully automated installation.

---

Step One: Install Fedora Linux 27 Server

That's it. Download the ISO from Fedora and do the most basic possible installation. This tool is designed to work with both Server and Minimal installation methods, but is primarily tested against Server. During installation simply be sure to set up networking during the installation. Once you are done installation, log in as root.

Step Two: Run the installation command...

source <(curl -s https://gitlab.com/scottalanmiller/nextcloud_fedora_installer/raw/master/nextcloud_fedora.sh)

When this command runs it will ask a few basic user and password questions. And that's it, just sit back and wait.

Step Three: Log in and use Nextcloud!

https://youripaddress/nextcloud/

That's really it. Install the OS, run the command, and done. The system will automatically install all needed packages, configure the database, apply passwords, install and properly configure caches, even run through the normal installation you would do on the web site. Everything, it's done. Your new system is updated, secured, configured, and ready to use. Log in with the username and password that you provided when you started the script.

---

Current Fedora Version Tested: Fedora 27
Current NextCloud Version Tested: NextCloud 13.0

---

Notes:

This installation is meant to be a production ready server and, as such, is meant to be headless. There is no local GUI installed and no need to do any local configuration. The web interface is fully configured and secured before use to be accessed by another web browser on your network. Once installed, only port 443 (HTTPS) is open to the network for NextCloud.

Because of the install style used here, this installer will work perfectly well anywhere from installing on a physical box, a VM running on any hypervisor, or installed to any cloud environment that supports Fedora Linux. This install script is highly flexible and portable.

In the current version, the system sets up /data directory as the location for data files. If you are an advanced user and will be creating a discrete data filesystem, you will need to create /data and mount that filesystem there prior to running the script. I do this myself and recommend XFS on LVM2; NextCloud recommends BtrFS.

This install only does MariaDB which is the best option for nearly all deployments. If you want to use SQLite or PostgreSQL, let me know, the script will need to be expanded for those options.

SELinux is not disabled here. All the extra security that it provides is included with full configuration done as part of the install.

SSL Certifications and Backups are not configured in this script, please consider those things once this has been completed.

---

The purpose of this installer is to make it as simple as possible for anyone to get up and running with a well crafted, by the book, enterprise grade NextCloud install. No corners cut, ready for solid production use. It is meant to make NextCloud accessible to everyone. If you have advanced or very specific needs for your NextCloud install, you should consider a manual install, but this script should serve as a guide and starting point to that process.

Special thanks to @JaredBusch for his NextCloud on Fedora guide as a starting point for this script.

I often find that the ACT of posting something makes me think about it and fix the issue. The issue was that I was zipping the folder that contained the files rather than making a "flat zip" file. It is working now.

UrBackup makes Windows backups super easy. Not only does UrBackup support Windows really well, but the components for it are in Chocolatey! It doesn't get much better than that. And to top it off, your first backup is automatic.

So, if we assume that you already have installed a UrBackup Server and you want to test your first Windows client to be backed up, it's this simple....

On a Windows machine, open your Command Prompt or PowerShell as the administrator and run...

choco install urbackup-client -y

Um, like, that's it. That's the whole thing.

Where is the configuration you ask? There isn't any needed. The client and server will discover each other automatically and they will kick off an initial backup right away to make sure that your system is protected as quickly as possible.

eta: link to urbackup site for additional information.

I'll be back in 10... just going into the server room to download something.

I had no idea that this existed until today. If you log into ScreenConnect normally from ChromeOS, you get nothing, not even the account management fields in SC. It's pretty disappointing. What it should do, though, is tell you that there is an alternative.

If you are no ChromeOS (not Chrome on something else, only ChromeOS) there is a browser extension just for ChromeOS that adds ScreenConnect functionality in the browser and works insanely well. Better, in fact, IMHO than the normal ScreenConnect client! The speed is near instant and super smooth. I'm really impressed.

So if you use ScreenConnect and wonder if ChromeOS is an option, it definitely is!

If you did the latest round of CentOS 7 updates (and presumably RHEL 7 updates) yesterday, and you run a different MTA (email server) than the system expects, such as Zimbra, then you may run into an update bug. CentOS has an update to Postfix which causes it to become enabled, even when it was previously disabled, modifying a running system. It then binds to 127.0.0.1:25. On a "normal" box, this is not an issue. But if you have another MTA, such as when you are running a Zimbra server that needs to bind its own MTA to 0.0.0.0:25, then you have a port conflict.

Likely, many people running CentOS 7 who did not want Postfix to be running now have it running but since the port is not generally exposed they are not aware that anything has changed. Still bad, but pretty minor. But if you are running an email server, this will stop emails from being received. With Zimbra, it will stop receiving, but will continue sending because it can still use the new MTA for outbound emails.

Fixing this is simple, discovering it is not. To fix, simply...

systemctl disable postfix

And then reboot to bring whatever system you have back to the way that it was.