Writing a Snippet Argument

A Snippet Argument provides the execution plan to the Snippet Framework and enables low-code/no-code collections.

Before writing a Snippet Argument, we must first understand how to collect the required data and ensure that step(s) exist for the required tasks. If a step does not exist, we determine if we can chain steps together to get the required result. After we know what steps to use, then we write the Snippet Argument for the given Syntax.

The most basic form of writing a Snippet Argument is through the low_code Syntax which we use for the following examples.

CLI

The following steps are typically used for CLI-based collections:

The following steps are typically used if the previous steps do not satisfy your use-case:

The following CLI examples require collecting data in this sequence:

Perform an SSH request using the ssh step

Convert the CLI output into a Python data structure

Select information from that data structure

Monitoring Zombie Processes

For this example, we ssh into a device and run a command to count the number of zombie process.

In order, to find the number of processes, we run the ps command as follows.

ps -elF

We run this command on an Oracle Linux 8 device and it produces the following output:

F S UID        PID  PPID  C PRI  NI ADDR SZ WCHAN    RSS PSR STIME TTY          TIME CMD
S root         1     0  0  80   0 - 53678 ep_pol  4708   5 Jan09 ?        08:50:11 /usr/lib/systemd/systemd --switched-root --system --deserialize 22
S root         2     0  0  80   0 -     0 kthrea     0   6 Jan09 ?        00:00:20 [kthreadd]
S root         4     2  0  60 -20 -     0 worker     0   0 Jan09 ?        00:00:00 [kworker/0:0H]
S root         6     2  0  80   0 -     0 smpboo     0   0 Jan09 ?        00:36:22 [ksoftirqd/0]
S root         7     2  0 -40   - -     0 smpboo     0   0 Jan09 ?        00:01:06 [migration/0]

Next, we use the jc step which is able to parse the ps command. The jc step transforms the CLI output into the following python data structure.

{0: {'addr': '-',
   'c': 0,
   'cmd': '/usr/lib/systemd/systemd --switched-root --system --deserialize '
          '22',
   'f': '4',
   'ni': '0',
   'pid': 1,
   'ppid': 0,
   'pri': '80',
   'psr': '2',
   'rss': 4708,
   's': 'S',
   'stime': 'Jan09',
   'sz': '53678',
   'time': '08:50:12',
   'tty': None,
   'uid': 'root',
   'wchan': 'ep_pol'}, ...

We receive a list of dictionaries where the key s contains the process state. The value Z in this field indicates a zombie process. Next, we use the jmespath selector to select all the zombie processes. Our query looks like the following:

"[?s == `Z`]"

From here, we find the number of zombie instances using the JMESpath length function.

"length([?s == `Z`])"

Note

The monitored device might not have any zombie processes, so to verify the functionality of the snippet argument, we can change the jmespath query to monitor Sleeping versus a Zombie state. The query to output the number of sleeping processes is length([?s == `S`]).

For this scenario, we use the following steps:

Steps for collection
Operation	Step
Perform a SSH request	ssh
Convert response from a string to a dictionary	jc
Select data from a dictionary	jmespath

Now that we know what steps to use, we need to find what arguments, if any, need to be supplied to the step.

Steps for collection
Operation	Step
Query Prometheus for `prometheus_http_requests_total`	query
Interpret results as a vector	result_type
Retrieve the handler and status code labels	labels

Steps for collection
Operation	Step
Query Prometheus for `http_request_duration_seconds_count`	query
Interpret results as a matrix	result_type
Select the average value of each	aggregation
Retrieve the status code associated with each value	labels

Writing a Snippet Argument

CLI

Monitoring Zombie Processes

Determining Step Arguments

ssh

jc

jmespath

Creating the Snippet Argument

Getting IcmpMsg data from /proc/net/snmp

Determining Step Arguments

ssh

parse_proc_net_snmp

jmespath

Creating the Snippet Argument

Parsing lines from lscpu

Determining Step Arguments

ssh

parse_line

jmespath

Creating the Snippet Argument

Prometheus

Best Practices

Obtaining HTTP API Endpoint Statistics

Determining Step Arguments

query

result_type

labels

Creating the Snippet Argument

Obtaining HTTP Request Response Time Statistics

Determining Step Arguments

query

result_type

aggregation

labels

Creating the Snippet Argument

REST

Collecting device names from an AIO SL1 System

Determining Step Arguments

http

json

jmespath

Creating the Snippet Argument

Collecting aligned organization for a device

Determining Step Arguments

http

json

jmespath

Creating the Snippet Argument

SNMP

Obtaining sysDescr

Determining Step Arguments

snmp

Creating the Snippet Argument

Obtaining interface names

Determining Step Arguments

snmp

Creating the Snippet Argument