Skip to content

Latest commit

 

History

History
100 lines (75 loc) · 5.48 KB

README.md

File metadata and controls

100 lines (75 loc) · 5.48 KB
Raw

argo-ams-consumer

Description

argo-ams-consumer component fetchs metric results from ARGO Messaging Service and writes them into avro serialized files. Files are written on daily basis and represent full daily log of metric results formed and sent from Nagios monitoring instances. It is functioning as unix service/daemon and it's multi-tenant aware meaning that multiple instances of it can be spawned on the same machine where each instance can be configured for specific tenant.

Component is mainly serving for redundancy/backup of metric results within ARGO framework in case of erroneous behaviour of Flink/HBase stack.

Installation

Component is supported on Centos 7. RPM packages and all needed dependencies are available in ARGO repositories so installation of component simply narrows down to installing a package:

yum install -y argo-ams-consumer

For its functioning, component depends on:

  • argo-ams-library - interaction with ARGO Messaging Service
  • avro - avro serialization of messages' payload
  • python-argparse - ease build and parse of command line arguments
  • python-daemon - ease daemonizing of component
File Types Destination
Configuration /etc/argo-ams-consumer/ams-consumer.conf
Avro Schema /etc/argo-ams-consumer/metric_data.avsc
Daemon component /usr/bin/ams-consumerd
Service unit /usr/lib/systemd/system/ams-consumer@.service
Target (units group) /usr/lib/systemd/system/ams-consumers.target
Pid files /var/run/argo-ams-consumer/
Log files /var/log/argo-ams-consumer/
Daily logs /var/lib/argo-ams-consumer/

Configuration

Configuration is done in one file ams-consumer.conf that is splitted in several sections, [General], [AMS], [MsgRetention], [Output].

General

[General]
LogName = argo-ams-consumer
ReportWritMsgEveryHours = 24
AvroSchema = /etc/argo-ams-consumer/metric_data.avsc
  • Logname defines name of argo-ams-consumer tenant instance appearing in logs
  • ReportWritMsgEveryHours for given number of hours instruct component to report how many messages it fetched. Number will be reported in log files.
  • AvroSchema defines the format of messages (fields and values)

AMS

[AMS]
Host = messaging-devel.argo.grnet.gr
Project = EGI
Token = EGITOKEN
Subscriptions = sub1
PullMsgs = 100
PullIntervalSec = 3
ConnectionTimeout = 180
PullRetry = 5
PullRetrySleep = 60
  • Host is FQDN of Argo Messaging Service instance
  • Project represents tenant within Argo Messaging Service
  • Subscription is the name of bucket containing metric results of interest
  • Token is authorization credential needed for pull operation on subscription
  • PullMsgs is a number of messages that will be fetched in single request
  • PullIntervalSec is interval between two pull requests and can be specified as fraction of second
  • ConnectionTimeout represents time window within that request must be served
  • PullRetry is number of retries for a single pull request that may fail because of connection problems and other hiccups that may occur
  • PullRetrySleep represents time in seconds after next pull will be tried

MsgRetention

Section defines the time window within the messages will be considered valid and fetched from subscription.

[MsgRetention]
PastDaysOk = 3
FutureDaysOk = 1
  • PastDaysOk defines the number of past days
  • FutureDaysOk defines the number of future days to consider

Output

[Output]
Directory = /var/lib/argo-ams-consumer
Filename = argo-consumer_log_DATE.avro
ErrorFilename = argo-consumer_error_log_DATE.avro
  • Directory is path to directory where avro files with metric results for given day will be stored
  • Filename is name of the avro file that will be created for each day. DATE is a placeholder in form of <year>_<month>_<day>
  • ErrorFilename is name of the avro file that will contain metric results fetched outside of allowed message retention defined in previous section

Configuration - multitenancy

Each tenant is represented with its own set of metric results meaning that AMS consumer needs to be configured with their specific subscription and different output directory where their data will be stored. Consumer is functioning as a unix service and multiple instances of it can be spawned if consumer is started with different config file, each tied to specific tenant. Name of pid file created for each instance of consumer is tied to filename of passed config and multiple instances of consumer with the same config file will be prevented. So if one wants to spawn a new consumer for new tenant, procedure is this:

  1. created separate config file /etc/argo-ams-consumer/ams-consumer-tenant.conf with changed settings in [AMS] and [Output] section at least:
    • change Project, Subscription, Token
    • change output directory to /var/lib/argo-ams-consumer-tenant/ that needs to be manually created before
  2. edit SystemD target that references and groups all units /usr/lib/systemd/system/ams-consumers.target by adding new unit in Requires=... ams-consumer@tenant.service ...
  3. reload SystemD configuration to register changes in target - systemctl daemon-reload
  4. start new unit with systemctl start ams-consumers.target