v1.3.6

What's Changed

• Support for FreeBSD by @dveeden in #63
• Add promote job to gosigar pipeline by @selzoc in #66

New Contributors

• @dveeden made their first contribution in #63
• @selzoc made their first contribution in #66

Full Changelog: v1.3.5...v1.3.6

v1.3.5

What's Changed

• /etc/mtab may not exist on containers, use /proc/mounts if not found by @digiserg in #60

New Contributors

• @digiserg made their first contribution in #60

Full Changelog: v1.3.4...v1.3.5

v1.3.4 - Correctly report maximum memory when running in nested cgroups.

Summary

This release contains PR #57. Reporting of maximum available memory for programs running in nested cgroups should now be correct.

For more information about the specifics of the issue, please consult this comment: #57 (comment).

Acknowledgements

Many thanks to @Nicholaswang for submitting PR #57 and bearing with us while we suggested changes to it.

v1.3.3 - No longer panic when encountering cgroup controllers mounted in multiple locations

Fixes Issue #54. If a cgroup controller is mounted in multiple locations, the first mountpoint will be used rather than panicking.

Full Changelog: v1.3.2...v1.3.3

v1.3.2 - Export the new `GetMemIgnoringCGroups` method properly

Properly export method as part of Sigar interface

[#179517901] Update Bosh Agent to correctly report memory used in `bosh vms --vitals` and related code paths, and publish new stemcells

Signed-off-by: Brian Upton <bupton@vmware.com>

v1.3.1 - Add GetMemIgnoringCGroups to FakeSigar for testing

Add GetMemIgnoringCGroups to fake sigar interface

[#179517901] Update Bosh Agent to correctly report memory used in `bosh vms --vitals` and related code paths, and publish new stemcells

Signed-off-by: Kenneth Lakin <klakin@vmware.com>

v1.3.0 - GetMemIgnoringCGroups

v1.2.0 changed the behavior of the GetMem method to return cgroup-constrained memory information. This may not be the desired behavior for all use cases. v1.3.0 introduces a new method GetMemIgnoringCGroups that return system level memory information instead.

v1.2.0: Query cgroup memory data and fuse with /proc data for more reliable information

Summary

This release augments the existing memory reporting functionality to detect when confined in a memory cgroup and report the cgroup limits, rather than the whole-system limits; if the cgroup limits are smaller than the whole-system limits.

For more information, check out PR #50.

Acknowledgements

@mudler for the initial debugging work that lead to identifying this deficiency as the root cause of log-cache failures in containerized environments: #48
@andreas-kupries for writing the implementation and tests
@jandubois for ensuring that this PR didn't get lost in the shuffle
@klakin-pivotal for PR review

v1.1.0: skip psnotify tests when running in container

[#152736258](https://www.pivotaltracker.com/story/show/152736258)