Marco Pracucci pracucci

Design

Web workers can use the Prometheus PHP client to track metrics. Such metrics are stored in memory and also copied to a shared memory from which another process in the same SysV IPC namespace can read. This design allows us to run a micro webserver in a sidecar container from which we can export Prometheus metrics collected in the worker's cli app.

IPC namespace: when deployed on K8S, two containers of the same Pod run in the same IPC namespace, while two different pods on the same host run on a different IPC namespace unless explicitly configured to all share the host's IPC running the pod with hostIPC: true.

Why this design

There are three options to export metrics to Prometheus from a cli app:

The halt command on EC2 Nitro instances doesn't poweroff the instance, so the OS will shutdown but the instance will be in the running state indefinitely (until you force a "stop" or "termination" via EC2 console / API). This is because the hypervisor has changed and the shutdown signaling between OS and the hypervisor behaves differently.

Works (the instance switches to "stopped" or "terminated"):

halt --poweroff
shutdown now (defaults to --poweroff)

Doesn't work (the instance hungs into a "running" state):

halt
shutdown --halt

	Cursor cursor = getActivity().getContentResolver().query(
	Uri.parse("content://com.google.android.music.MusicContent/audio"),
	new String[] { "_id", "CanonicalName" }, // new String[]{"_id", "playlist_name"},
	null, null, null);

	// android.database.sqlite.SQLiteException: no such column: SongCount (code 1): , while compiling: SELECT null AS containerType, CanonicalAlbum, CanonicalName, max(Nid) AS Nid, Album AS album, (MIN(LocalCopyType) != 300) AS hasRemote, null AS domainParam, max(StoreAlbumId) AS StoreAlbumId, Composer AS composer, 'audio/*' AS mime_type, count(distinct(SongId)) AS _count, SongCount AS keeponSongCount, Genre AS Genre, Year AS year, Size, 0 AS bookmark, Domain, 1 AS is_music, DiscNumber, MUSIC.SourceId AS SourceId, FileDate, CanonicalAlbumArtist AS artistSort, Rating, null AS containerName, BitRate, null AS containerId, MUSIC.SourceAccount AS SourceAccount, AlbumArtist, AlbumArtistId, (MAX(LocalCopyType) I

	/**
	* Returns the RGB color of a string.
	*/
	public static int getColorFromString(String input)
	{
	int hash = 0;
	int prc = -10;

	// Generate hash
	for (int i = 0; i < input.length(); i++) {

	#!/bin/bash

	# Init
	OPT_FROM_FILES=
	OPT_DEPLOYMENT=
	OPT_CONTAINER=
	OPT_SIGNAL=
	OPT_YES=0

	# Helper to print usage

	FROM ubuntu:16.04

	# Config
	ARG SBT_VERSION=0.13.12
	ARG SJS_VERSION=0.8.0
	ARG EMR_VERSION=5.13.0
	ARG SPARK_VERSION=2.3.0

	# Install JDK 8 and some dependencies
	RUN apt-get update -qq && apt-get install -y -qq openjdk-8-jdk wget python python-pip

	#!/bin/bash
	#
	# This is the script run on master as bootstrap action.
	#

	set -e

	# Config
	SJS_VERSION=0.8.0
	EMR_VERSION=5.13.0

	#!/bin/bash
	#
	# On 2018-08-21 we found out the existance of a bug in the Kubernetes cluster
	# which leaves stale / dangling cgroups on the system related to secret volume
	# mounts.
	#
	# The root cause of this bug is not clear yet, even if it's likely to be a systemd
	# bug because the kubelet runs `systemd-run` command to mount secret volumes and,
	# according to logs, successfully `unmount` such volumes once the pods terminates.
	#