2eff100d5d
* Remove duplicate partial reference from release-notes (#24081) * remove partial references from release-notes that link to upgrade guides, and change link in partial to anchor * Clarify leak is memory consumption There is no leak of information. * update references in table * update table to include range for affected versions --------- Co-authored-by: Meggie Ladlow <meggie@hashicorp.com> * update link in known issues table --------- Co-authored-by: davidadeleon <56207066+davidadeleon@users.noreply.github.com> Co-authored-by: Meggie Ladlow <meggie@hashicorp.com>
23 lines
893 B
Plaintext
23 lines
893 B
Plaintext
### Fatal error during expiration metrics gathering causing Vault crash
|
|
|
|
#### Affected versions
|
|
|
|
This issue affects Vault Community and Enterprise versions:
|
|
- 1.13.9
|
|
- 1.14.5
|
|
- 1.15.1
|
|
|
|
A fix has been issued in Vault 1.13.10, 1.14.6, and 1.15.2.
|
|
|
|
#### Issue
|
|
|
|
A recent change to Vault to improve state change speed (e.g. becoming active or standby) introduced a concurrency issue
|
|
which can lead to a concurrent iteration and write on a map, causing a fatal error and crashing Vault. This error occurs
|
|
when gathering lease and token metrics from the expiration manager. These metrics originate from the active node in a HA
|
|
cluster, as such a standby node will take over active duties and the cluster will remain functional should the original
|
|
active node encounter this bug. The new active node will be vulnerable to the same bug, but may not encounter it immediately.
|
|
|
|
There is no workaround.
|
|
|
|
|