Google Cloud Platform Condor Pool Manager (GCPM)

HTCondor pool manager for Google Cloud Platform.

Installation

Package installation

GCPM can be installed by pip:

$ pip install gcpm

Service file installation

To install as service, do:

$ gcpm install

⚠️ Service installation is valid only for the system managed by Systemd.

If logrotate is installed, logrotation definition for /var/log/gcpm.log is also installed.

Configuration file

Configuration file path

The default configuration file is ~/.config/gcpm/gcpm.yml.

For service, the configuration file is /etc/gcpm.yml.

To change the configuration file, use --config option:

$ gcpm run --config /path/to/my/gcpm.yml

Configuration file content

A configuration file is YAML format.

Name	Description	Default Value	Mandatory
config_dir	Directory for some gcpm related files.	~/.config/gcpm/ (user) /var/cache/gcpm (service)	No
oatuh_file	Path to OAuth information file for GCE/GCS usage.	<config_dir>/oauth	No
service_account_file	Service account JSON file for GCE/GCS usage. If not specified, OAuth connection is tried.	-	No
project	Google Cloud Platform Project Name.	-	Yes
zone	Zone for Google Compute Engine.	-	Yes
machines	Array of machine settings. Each setting is array of [core, mem, disk, idle, image] (see below).	[]	Yes
machines:core	Number of core of the machine type.	-	Yes
machines:mem	Memory (MB) of the machine type.	-	Yes
machines:disk	Disk size (GB) of the machine type.	-	Yes
machines:max	Limit of the number of instances for the machine type.	-	Yes
machines:idle	Number of idle machines for the machine type.	-	Yes
machines:image	Image of the machine type.	-	Yes
machines:<others>	Other any options can be defined for creating instance.	-	No
max_cores	Limit of the total number of cores of all instances. If it is set 0, no limit is applied.	0	No
static_wns	Array of instance names of static worker nodes, which are added as condor worker nodes.	[]	No
required_machines	Array of machines which should be running other than worker nodes.	[]	No
required_machines:name	Number of core of the machine type.	-	Yes
required_machines:mem	Memory (MB) of the machine type.	-	Yes
required_machines:disk	Disk size (GB) of the machine type.	-	Yes
required_machines:image	Image of the machine type.	-	Yes
required_machines:<others>	Other any options can be defined for creating instance.	-	No
primary_accounts	User accounts which jobs must run normal worker nodes. See below about primary accounts.	[]	No
prefix	Prefix of machine names.	gcp-wn	No
preemptible	1 for preemptible machines, 0 for not.	0	No
off_timer	Second to send condor_off after starting.	0	No
network_tag	Array of GCP network tag.	[]	No
reuse	1 to reused terminated instance. Otherwise delete and re-created instances.	0	No
interval	Second of interval for each loop.	10	No
head_info	If head is empty, head node information is automatically taken for each option: hostname: Hostname ip: IP address gcp: Hostname	gcp	No
head	Head node Hostname/IP address.	""	No
port	HTCondor port.	9618	No
domain	Domain of the head node. Set empty to take it from hostnaem.	""	No
admin	HTCondor admin email address.	""	Yes
owner	HTCondor owner name.	""	Yes
wait_cmd	1 to wait GCE commands result (create/start/stop/delete...).	0	No
bucket	Bucket name for pool_password file.	""	Yes
storageClass	Storage class name of the bucket.	"REGIONAL"	No
location	Storage location for the bucket. If empty, it is decided from the zone.	""	No
log_file	Log file path. Empty to put it in stdout.	""	No
log_level	Log level. (debug, info, warning, error, critical)	info	No

Note:

Primary accounts

If primary accounts are set, jobs of non-primary accounts can run on test worker nodes.

If there are already max number of 1 core worker nodes and idle jobs of non-primary accounts are there, test worker node named <prefix&gt-test-1core-XXXX will be launched and only non-primary account jobs can run on it.

This able to run such a test job w/o waiting for finishing any normal jobs.

Such test worker nodes can be launched until total cores are smaller than max_core.

To use this function effectively, set total of max of each core to less than max_core.

e.g.)

---
machines:
  core: 1
  max: 10
machines:
  core: 8
  max:  2
max_core: 20
primary_accounts:
  - condor_primary

In this case, normal jobs can launch 10 1 core machines and 2 8 core machines, then 16 cores are used.

Even if there are a log of idle condor_primary's jobs, 1 core test jobs by other accounts can run: 4 jobs at most.

Metadata

Release files for gcpm 0.2.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for gcpm 0.2.0
File	Size	Uploaded
gcpm-0.2.0.tar.gz	19.9 kB	Jan 20, 2019	Details

Built distribution (wheel)

Table of built distributions (wheels) for gcpm 0.2.0
File	Interpreter	ABI	Platform	Reset
gcpm-0.2.0-py2.py3-none-any.whl 67.0 kB Jan 20, 2019	Python 2, Python 3	none	any	Details

Total release size: 87.0 kB

Release files / gcpm-0.2.0.tar.gz

Download URL	gcpm-0.2.0.tar.gz
Size	19.9 kB
Tags	Source
SHA-256 checksum How to use checksums	`225531c239c1f4c8a4f7b37f195aec31f811b430b9d05bc3413f8259a346ab81`
BLAKE2b-256 checksum How to use checksums	`c5dfe426ac067d6440bbd695baae58455efccb9e96962fe13fed65792474dc60`
Upload date	Jan 20, 2019
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`poetry/0.12.10 CPython/2.7.15 Darwin/18.2.0`

Release files / gcpm-0.2.0-py2.py3-none-any.whl

Download URL	gcpm-0.2.0-py2.py3-none-any.whl
Size	67.0 kB
Tags	Python 2 Python 3
SHA-256 checksum How to use checksums	`7175fe437e68e0822e4f3305f585f990eb749c95070b23c23625b30655184a5d`
BLAKE2b-256 checksum How to use checksums	`28c37d86b3ab38f416306c2007ee897ff8f9e467dfc69751218d34a6c1b2c941`
Upload date	Jan 20, 2019
Uploaded using Trusted Publishing? What is trusted publishing?	No
Uploaded via	`poetry/0.12.10 CPython/2.7.15 Darwin/18.2.0`