README.md

apps.plugin

apps.plugin breaks down system resource usage to processes, users and user groups.

To achieve this task, it iterates through the whole process tree, collecting resource usage information for every process found running.

Since Netdata needs to present this information in charts and track them through time, instead of presenting a top like list, apps.plugin uses a pre-defined list of process groups to which it assigns all running processes. This list is customizable via apps_groups.conf, and Netdata ships with a good default for most cases (to edit it on your system run /etc/netdata/edit-config apps_groups.conf).

So, apps.plugin builds a process tree (much like ps fax does in Linux), and groups processes together (evaluating both child and parent processes) so that the result is always a list with a predefined set of members (of course, only process groups found running are reported).

If you find that apps.plugin categorizes standard applications as other, we would be glad to accept pull requests improving the defaults shipped with Netdata in apps_groups.conf.

Unlike traditional process monitoring tools (like top), apps.plugin is able to account the resource utilization of exit processes. Their utilization is accounted at their currently running parents. So, apps.plugin is perfectly able to measure the resources used by shell scripts and other processes that fork/spawn other short lived processes hundreds of times per second.

Charts

apps.plugin provides charts for 3 sections:

1. Per application charts as Applications at Netdata dashboards
2. Per user charts as Users at Netdata dashboards
3. Per user group charts as User Groups at Netdata dashboards

Each of these sections provides the same number of charts:

• CPU utilization (apps.cpu)
  • Total CPU usage
  • User/system CPU usage (apps.cpu_user/apps.cpu_system)
• Disk I/O
  • Physical reads/writes (apps.preads/apps.pwrites)
  • Logical reads/writes (apps.lreads/apps.lwrites)
  • Open unique files (if a file is found open multiple times, it is counted just once, apps.files)
• Memory
  • Real Memory Used (non-shared, apps.mem)
  • Virtual Memory Allocated (apps.vmem)
  • Minor page faults (i.e. memory activity, apps.minor_faults)
• Processes
  • Threads running (apps.threads)
  • Processes running (apps.processes)
  • Carried over uptime (since the last Netdata Agent restart, apps.uptime)
  • Minimum uptime (apps.uptime_min)
  • Average uptime (apps.uptime_average)
  • Maximum uptime (apps.uptime_max)
  • Pipes open (apps.pipes)
• Swap memory
  • Swap memory used (apps.swap)
  • Major page faults (i.e. swap activity, apps.major_faults)
• Network
  • Sockets open (apps.sockets)

In addition, if the eBPF collector is running, your dashboard will also show an additional list of charts using low-level Linux metrics.

The above are reported:

• For Applications per target configured.
• For Users per username or UID (when the username is not available).
• For User Groups per groupname or GID (when groupname is not available).

Performance

apps.plugin is a complex piece of software and has a lot of work to do We are proud that apps.plugin is a lot faster compared to any other similar tool, while collecting a lot more information for the processes, however the fact is that this plugin requires more CPU resources than the netdata daemon itself.

Under Linux, for each process running, apps.plugin reads several /proc files per process. Doing this work per-second, especially on hosts with several thousands of processes, may increase the CPU resources consumed by the plugin.

In such cases, you many need to lower its data collection frequency.

To do this, edit /etc/netdata/netdata.conf and find this section:

[plugin:apps]
	# update every = 1
	# command options =

Uncomment the line update every and set it to a higher number. If you just set it to 2, its CPU resources will be cut in half, and data collection will be once every 2 seconds.

Configuration

The configuration file is /etc/netdata/apps_groups.conf. To edit it on your system, run /etc/netdata/edit-config apps_groups.conf.

The configuration file works accepts multiple lines, each having this format:

group: process1 process2 ...

Each group can be given multiple times, to add more processes to it.

For the Applications section, only groups configured in this file are reported. All other processes will be reported as other.

For each process given, its whole process tree will be grouped, not just the process matched. The plugin will include both parents and children. If including the parents into the group is undesirable, the line other: * should be appended to the apps_groups.conf.

The process names are the ones returned by:

• ps -e or cat /proc/PID/stat
• in case of substring mode (see below): /proc/PID/cmdline

To add process names with spaces, enclose them in quotes (single or double) example: 'Plex Media Serv' or "my other process".

You can add an asterisk * at the beginning and/or the end of a process:

• *name suffix mode: will search for processes ending with name (at /proc/PID/stat)
• name* prefix mode: will search for processes beginning with name (at /proc/PID/stat)
• *name* substring mode: will search for name in the whole command line (at /proc/PID/cmdline)

If you enter even just one name (substring), apps.plugin will process /proc/PID/cmdline for all processes (of course only once per process: when they are first seen).

To add processes with single quotes, enclose them in double quotes: "process with this ' single quote"

To add processes with double quotes, enclose them in single quotes: 'process with this " double quote'

If a group or process name starts with a -, the dimension will be hidden from the chart (cpu chart only).

If a process starts with a +, debugging will be enabled for it (debugging produces a lot of output - do not enable it in production systems).

You can add any number of groups. Only the ones found running will affect the charts generated. However, producing charts with hundreds of dimensions may slow down your web browser.

The order of the entries in this list is important: the first that matches a process is used, so put important ones at the top. Processes not matched by any row, will inherit it from their parents or children.

The order also controls the order of the dimensions on the generated charts (although applications started after apps.plugin is started, will be appended to the existing list of dimensions the netdata daemon maintains).

There are a few command line options you can pass to apps.plugin. The list of available options can be acquired with the --help flag. The options can be set in the netdata.conf file. For example, to disable user and user group charts you should set

[plugin:apps]
  command options = without-users without-groups

Integration with eBPF

If you don't see charts under the eBPF syscall or eBPF net sections, you should edit your ebpf.d.conf file to ensure the eBPF program is enabled.

Also see our guide on troubleshooting apps with eBPF metrics for ideas on how to interpret these charts in a few scenarios.

Permissions

apps.plugin requires additional privileges to collect all the information it needs. The problem is described in issue #157.

When Netdata is installed, apps.plugin is given the capabilities cap_dac_read_search,cap_sys_ptrace+ep. If this fails (i.e. setcap fails), apps.plugin is setuid to root.

linux capabilities in containers

There are a few cases, like docker and virtuozzo containers, where setcap succeeds, but the capabilities are silently ignored (in lxc containers setcap fails).

In these cases ()setcap succeeds but capabilities do not work), you will have to setuid to root apps.plugin by running these commands:

chown root:netdata /usr/libexec/netdata/plugins.d/apps.plugin
chmod 4750 /usr/libexec/netdata/plugins.d/apps.plugin

You will have to run these, every time you update Netdata.

Security

apps.plugin performs a hard-coded function of building the process tree in memory, iterating forever, collecting metrics for each running process and sending them to Netdata. This is a one-way communication, from apps.plugin to Netdata.

So, since apps.plugin cannot be instructed by Netdata for the actions it performs, we think it is pretty safe to allow it have these increased privileges.

Keep in mind that apps.plugin will still run without escalated permissions, but it will not be able to collect all the information.

Application Badges

You can create badges that you can embed anywhere you like, with URLs like this:

https://your.netdata.ip:19999/api/v1/badge.svg?chart=apps.processes&dimensions=myapp&value_color=green%3E0%7Cred

The color expression unescaped is this: value_color=green>0|red.

Here is an example for the process group sql at https://registry.my-netdata.io:

Netdata is able give you a lot more badges for your app. Examples below for process group sql:

• CPU usage:
• Disk Physical Reads
• Disk Physical Writes
• Disk Logical Reads
• Disk Logical Writes
• Open Files
• Real Memory
• Virtual Memory
• Swap Memory
• Minor Page Faults
• Processes
• Threads
• Major Faults (swap activity)
• Open Pipes
• Open Sockets

For more information about badges check Generating Badges

Comparison with console tools

SSH to a server running Netdata and execute this:

while true; do ls -l /var/run >/dev/null; done

In most systems /var/run is a tmpfs device, so there is nothing that can stop this command from consuming entirely one of the CPU cores of the machine.

As we will see below, none of the console performance monitoring tools can report that this command is using 100% CPU. They do report of course that the CPU is busy, but they fail to identify the process that consumes so much CPU.

Here is what common Linux console monitoring tools report:

top

top reports that bash is using just 14%.

If you check the total system CPU utilization, it says there is no idle CPU at all, but top fails to provide a breakdown of the CPU consumption in the system. The sum of the CPU utilization of all processes reported by top, is 15.6%.

top - 18:46:28 up 3 days, 20:14,  2 users,  load average: 0.22, 0.05, 0.02
Tasks:  76 total,   2 running,  74 sleeping,   0 stopped,   0 zombie
%Cpu(s): 32.8 us, 65.6 sy,  0.0 ni,  0.0 id,  0.0 wa,  1.3 hi,  0.3 si,  0.0 st
KiB Mem :  1016576 total,   244112 free,    52012 used,   720452 buff/cache
KiB Swap:        0 total,        0 free,        0 used.   753712 avail Mem

  PID USER      PR  NI    VIRT    RES    SHR S %CPU %MEM     TIME+ COMMAND
12789 root      20   0   14980   4180   3020 S 14.0  0.4   0:02.82 bash
    9 root      20   0       0      0      0 S  1.0  0.0   0:22.36 rcuos/0
  642 netdata   20   0  132024  20112   2660 S  0.3  2.0  14:26.29 netdata
12522 netdata   20   0    9508   2476   1828 S  0.3  0.2   0:02.26 apps.plugin
    1 root      20   0   67196  10216   7500 S  0.0  1.0   0:04.83 systemd
    2 root      20   0       0      0      0 S  0.0  0.0   0:00.00 kthreadd

htop

Exactly like top, htop is providing an incomplete breakdown of the system CPU utilization.

  CPU[||||||||||||||||||||||||100.0%]   Tasks: 27, 11 thr; 2 running
  Mem[||||||||||||||||||||85.4M/993M]   Load average: 1.16 0.88 0.90
  Swp[                         0K/0K]   Uptime: 3 days, 21:37:03

  PID USER      PRI  NI  VIRT   RES   SHR S CPU% MEM%   TIME+  Command
12789 root       20   0 15104  4484  3208 S 14.0  0.4 10:57.15 -bash
 7024 netdata    20   0  9544  2480  1744 S  0.7  0.2  0:00.88 /usr/libexec/netd
 7009 netdata    20   0  138M 21016  2712 S  0.7  2.1  0:00.89 /usr/sbin/netdata
 7012 netdata    20   0  138M 21016  2712 S  0.0  2.1  0:00.31 /usr/sbin/netdata
  563 root	     20   0  308M  202M  202M S  0.0 20.4  1:00.81 /usr/lib/systemd/
 7019 netdata    20   0  138M 21016  2712 S  0.0  2.1  0:00.14 /usr/sbin/netdata

atop

atop also fails to break down CPU usage.

ATOP - localhost            2016/12/10  20:11:27    -----------      10s elapsed
PRC | sys    1.13s | user   0.43s | #proc     75 | #zombie    0 | #exit   5383 |
CPU | sys      67% | user     31% | irq       2% | idle      0% | wait      0% |
CPL | avg1    1.34 | avg5    1.05 | avg15   0.96 | csw    51346 | intr   10508 |
MEM | tot   992.8M | free  211.5M | cache 470.0M | buff   87.2M | slab  164.7M |
SWP | tot     0.0M | free    0.0M |              | vmcom 207.6M | vmlim 496.4M |
DSK |          vda | busy      0% | read       0 | write      4 | avio 1.50 ms |
NET | transport    | tcpi      16 | tcpo      15 | udpi       0 | udpo       0 |
NET | network      | ipi       16 | ipo       15 | ipfrw      0 | deliv     16 |
NET | eth0    ---- | pcki      16 | pcko      15 | si    1 Kbps | so    4 Kbps |

  PID SYSCPU USRCPU   VGROW  RGROW  RDDSK   WRDSK ST EXC  S  CPU CMD       1/600
12789  0.98s  0.40s      0K     0K     0K    336K --   -  S  14% bash
    9  0.08s  0.00s      0K     0K     0K      0K --   -  S   1% rcuos/0
 7024  0.03s  0.00s      0K     0K     0K      0K --   -  S   0% apps.plugin
 7009  0.01s  0.01s	     0K     0K     0K      4K --   -  S   0% netdata

glances

And the same is true for glances. The system runs at 100%, but glances reports only 17% per process utilization.

Note also, that being a python program, glances uses 1.6% CPU while it runs.

localhost                                               Uptime: 3 days, 21:42:00

CPU  [100.0%]   CPU     100.0%   MEM     23.7%   SWAP      0.0%   LOAD    1-core
MEM  [ 23.7%]   user:    30.9%   total:   993M   total:       0   1 min:    1.18
SWAP [  0.0%]   system:  67.8%   used:    236M   used:        0   5 min:    1.08
                idle:     0.0%   free:    757M   free:        0   15 min:   1.00

NETWORK     Rx/s   Tx/s   TASKS  75 (90 thr), 1 run, 74 slp, 0 oth
eth0        168b    2Kb
eth1          0b     0b     CPU%  MEM%   PID USER        NI S Command
lo            0b     0b     13.5   0.4 12789 root         0 S -bash
                             1.6   2.2  7025 root         0 R /usr/bin/python /u
DISK I/O     R/s    W/s      1.0   0.0     9 root         0 S rcuos/0
vda1           0     4K      0.3   0.2  7024 netdata      0 S /usr/libexec/netda
                             0.3   0.0     7 root         0 S rcu_sched
FILE SYS    Used  Total      0.3   2.1  7009 netdata      0 S /usr/sbin/netdata
/ (vda1)   1.56G  29.5G      0.0   0.0    17 root         0 S oom_reaper

why does this happen?

All the console tools report usage based on the processes found running at the moment they examine the process tree. So, they see just one ls command, which is actually very quick with minor CPU utilization. But the shell, is spawning hundreds of them, one after another (much like shell scripts do).

What does Netdata report?

The total CPU utilization of the system:

Figure 1: The system overview section at Netdata, just a few seconds after the command was run

And at the applications apps.plugin breaks down CPU usage per application:

Figure 2: The Applications section at Netdata, just a few seconds after the command was run

So, the ssh session is using 95% CPU time.

Why ssh?

apps.plugin groups all processes based on its configuration file. The default configuration has nothing for bash, but it has for sshd, so Netdata accumulates all ssh sessions to a dimension on the charts, called ssh. This includes all the processes in the process tree of sshd, including the exited children.

Distributions based on systemd, provide another way to get cpu utilization per user session or service running: control groups, or cgroups, commonly used as part of containers apps.plugin does not use these mechanisms. The process grouping made by apps.plugin works on any Linux, systemd based or not.

a more technical description of how Netdata works

Netdata reads /proc/<pid>/stat for all processes, once per second and extracts utime and stime (user and system cpu utilization), much like all the console tools do.

But it also extracts cutime and cstime that account the user and system time of the exit children of each process. By keeping a map in memory of the whole process tree, it is capable of assigning the right time to every process, taking into account all its exited children.

It is tricky, since a process may be running for 1 hour and once it exits, its parent should not receive the whole 1 hour of cpu time in just 1 second - you have to subtract the cpu time that has been reported for it prior to this iteration.

It is even trickier, because walking through the entire process tree takes some time itself. So, if you sum the CPU utilization of all processes, you might have more CPU time than the reported total cpu time of the system. Netdata solves this, by adapting the per process cpu utilization to the total of the system. Netdata adds charts that document this normalization.