What It Does¶
- Check current AWS resource usage against AWS Service Limits
- Show and inspect current usage
- Override default Service Limits (for accounts with increased limits)
- Compare current usage to limits; return information about limits that exceed thresholds, and (CLI wrapper) exit non-0 if thresholds are exceeded
- Define custom thresholds per-limit
- Where possible, pull current limits from Trusted Advisor API
- Supports explicitly setting the AWS region
- Supports using STS
to assume roles in other accounts, including using
- Optionally refresh Trusted Advisor “Service Limits” check before polling Trusted Advisor data, and optionally wait for the refresh to complete (up to an optional maximum time limit). See Getting Started - Trusted Advisor for more information.
- An AWS Service or Product, such as EC2, VPC, RDS or ElastiCache. More specifically, Services in AwsLimitChecker correspond to distinct APIs for AWS Services.
- An AWS-imposed maximum usage for a certain resource type in AWS. See AWS Service Limits.
Limits are generally either account-wide or per-region. They have AWS global default values, but can be increased by AWS Support. “Limit” is also the term used
within this documentation to describe
AwsLimitobjects, which describe a specific AWS Limit within this program.
- “Usage” refers to your current usage of a specific resource that has a limit. Usage values/amounts (some integer or floating point number, such as number of VPCs
or GB of IOPS-provisioned storage) are represented by instances of the
AwsLimitUsageclass. Limits that are measured as a subset of some “parent” resource, such as “Subnets per VPC” or “Read Replicas per Master” have their usage tracked per parent resource, so you can easily determine which ones are problematic.
- The point at which AwsLimitChecker will consider the current usage for a limit to be problematic. Global thresholds default to usage >= 80% of limit for “warning” severity,
and usage >= 99% of limit for “critical” severity. Limits which have reached or exceeded their threshold will be reported separately for warning and critical (we generally
consider “warning” to be something that will require human intervention in the near future, and “critical” something that is an immediate problem, i.e. should block
automated processes). The
awslimitcheckercommand line wrapper can override the default global thresholds. The
AwsLimitCheckerclass can both override global percentage thresholds, as well as specify per-limit thresholds as a percentage, a fixed usage value, or both. For more information on overriding thresholds, see Python Usage / Setting a Threshold Override as well as the documentation for
It’s recommended that you install into a virtual environment (virtualenv / venv). See the virtualenv usage documentation for more details, but the gist is as follows (the virtualenv name, “limitchecker” here, can be whatever you want):
virtualenv limitchecker source limitchecker/bin/activate pip install awslimitchecker
If you’re using awslimitchecker in automated tooling that recreates the virtualenv
(such as Jenkins or cron jobs, etc) you’ll probably want to install a specific version
so that the job doesn’t unexpectedly break. It’s recommended that you pin your installation
major version. According to awslimitchecker’s versioning policy,
this should ensure that you get the latest awslimitchecker version that’s compatible with
your IAM policy and dependencies and has no backwards-incompatible API changes.
Aside from STS, awslimitchecker does nothing with AWS credentials, it leaves that to boto itself.
You must either have your credentials configured in one of boto3’s supported config
files or set as environment variables. If your credentials are in the cross-SDK
credentials file (
~/.aws/credentials) under a named profile section, you can
use credentials from that profile by specifying the
lint option. See
this project’s documentation
for further information.
Please note that version 0.3.0 of awslimitchecker moved from using
boto as its AWS API client to using
boto3. This change is mostly transparent, but there is a minor change in how AWS credentials are handled. In
boto, if the
AWS_SECRET_ACCESS_KEY environment variables were set, and the
region was not set explicitly via awslimitchecker, the AWS region would either be taken from the
environment variable or would default to us-east-1, regardless of whether a configuration file (
~/.aws/config) was present. With boto3, it appears that the default region from the configuration file will be
used if present, regardless of whether the credentials come from that file or from environment variables.
When using STS, you will need to specify the
--region option as well as the
--sts-account-role options to specify the Account ID that you want to assume a role in, and the
name of the role you want to assume. If an external ID is required, you can specify it with
In addition, when assuming a role STS, you can use a MFA device. simply
specify the device’s serial number with the
--mfa-serial-number option and a token generated by the device
--mfa-token option. STS credentials will be cached for the lifetime of the program.
Important Note on Session and Federation (Temporary) Credentials: The temporary credentials granted by the AWS IAM
API calls will throw errors when trying to access the IAM API (except for Session tokens, which will
work for IAM API calls only if an MFA token is used). Furthermore, Federation tokens cannot make use
of the STS AssumeRole functionality. If you attempt to use awslimitchecker with credentials generated
by these APIs (commonly used by organizations to hand out limited-lifetime credentials), you will likely
encounter errors when checking IAM limits. If this is acceptable, you can use these credentials by setting
AWS_SESSION_TOKEN environment variable in addition to
or by otherwise configuring these credentials in a way that’s supported by
To specify the region that
awslimitchecker connects to, use the
command line option. At this time awslimitchecker can only connect to one region at a time;
to check limits in multiple regions, simply run the script multiple times, once per region.
awslimitchecker supports retrieving your current service limits via the
“Service Limits” performance check
, for limits which Trusted Advisor tracks (currently a subset of what awslimitchecker
knows about). The results of this check may not be available via the API for all
accounts; as of December 2016, the Trusted Advisor documentation states that while
this check is available for all accounts, API access is only available to accounts
with Business- or Enterprise-level support plans. If your account does not have
Trusted Advisor access, the API call will result in a
and awslimitchecker will log a
Cannot check TrustedAdvisor message at
Trusted Advisor information is important to awslimitchecker, however, as it provides
the current service limit values for a number of limits that cannot be obtained
any other way. While you can completely disable Trusted Advisor polling via the
--skip-ta command-line option, you will then be left with default service
limit values for many limits.
As of 0.7.0, awslimitchecker also supports programmatically refreshing the “Service Limits” Trusted Advisor check, in order to get updated limit values. If this is not done, the data provided by Trusted Advisor may not be updated unless a human does so via the AWS Console. The refresh logic operates in one of three modes, controlled by command-line options (these are also exposed in the Python API; see the “Internals” link below):
--ta-refresh-wait- The check will be refreshed and awslimitchecker will poll every 30 seconds waiting for the refresh to complete (or until
ta_refresh_timeoutseconds have elapsed).
--ta-refresh-older INTEGER- This operates like the
--ta-refresh-waitoption, but will only refresh the check if its current result data is at least
--ta-refresh-trigger- The check will be refreshed and the program will continue on immediately, without waiting for the refresh to complete; this will almost certainly result in stale check results in the current run. However, this may be useful if you desire to keep
awslimitcheckerruns short, and run it on a regular schedule (i.e. if you run
awslimitcheckerevery 6 hours, and are OK with Trusted Advisor check data being 6 hours old).
Additionally, there is a
--ta-refresh-timeout option. If this is set (to an integer),
refreshes of the check will time out after that number of seconds. If a timeout
occurs, a message will be logged at error level, but the program will continue
running (most likely using the old result data).
Important: It may take 30 to 60 minutes for the Service Limits check to refresh on large accounts. Please be aware of this when enabling the refresh options.
Using the check refresh options will require the
See Internals - Trusted Advisor for technical information on the implementation of Trusted Advisor polling.
You can view a sample IAM policy listing the permissions required for awslimitchecker to function properly either via the CLI client:
Or as a python dict:
from awslimitchecker.checker import AwsLimitChecker c = AwsLimitChecker() iam_policy = c.get_required_iam_policy()
You can also view the required permissions for the current version of awslimitchecker at Required IAM Permissions.