OpenStack service logs missing in Elasticsearch
Bug #1930867 reported by
Doug Szumski
This bug affects 2 people
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
kolla |
Fix Released
|
High
|
Unassigned | ||
Victoria |
Fix Released
|
High
|
Unassigned | ||
Wallaby |
Fix Released
|
High
|
Unassigned | ||
Xena |
Fix Released
|
High
|
Unassigned |
Bug Description
As reported by Bernd Bausch on the mailing list there is an issue with forwarding logs to Elasticsearch or the Monasca API.
We see some logs arrive. When Fluentd is restarted we see the missing logs arrive in a burst, as if the local buffers have been flushed. An example
is apache_access logs, although people have also reported missing openstack service logs.
The issue is related to an upgrade of the td-agent package.
Fluentd 1.11.2 (comes via td-agent-
Fluentd 1.12.2/
Currently this package is not pinned.
description: | updated |
description: | updated |
Changed in kolla-ansible: | |
status: | New → Invalid |
no longer affects: | kolla-ansible |
Changed in kolla: | |
status: | New → In Progress |
To post a comment you must log in.
More specifically I can upgrade to Fluentd release 1.11.5 before I see this issue.
``` [root@control01 /]# td-agent-gem install fluentd -v 1.11.5
(fluentd)
```
The issue appears in 1.12.0 and doesn't go away in 1.13.0.
Some OpenStack service logs still arrive without changing the match tags from *.** to **.