Ceilometer api requests periodically fails with 'Cannot connect to proxy.', error(104, 'Connection reset by peer'
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
Fuel for OpenStack |
Fix Released
|
High
|
Ivan Berezovskiy | ||
8.0.x |
Fix Released
|
High
|
Ivan Berezovskiy | ||
Mitaka |
Fix Released
|
High
|
Ivan Berezovskiy |
Bug Description
See the bug https:/
Scenario:
1. Create new environment
2. Choose Neutron, VLAN
3. Choose Ceph for volumes and images, ceph for ephemeral and Rados GW for objects
4. Add 3 controller
5. Add 1 compute
6. Add 3 mongo nodes
7. Verify networks
8. Deploy cluster
9. Verify networks
10. Run OSTF
Multiple Ceilometer OSTF failed. In ostf.log we can see the following errors:
ConnectionError: HTTPSConnection
At a certain time requests do not reach to HAProxy.
We added in ceilometer haproxy config (/etc/haproxy/
option httplog
option httpclose
and all works fine.
I propose to add these options in the ceilometer haproxy config by default.
Changed in fuel: | |
assignee: | nobody → Ivan Berezovskiy (iberezovskiy) |
status: | New → Confirmed |
tags: | added: on-verification |
Changed in fuel: | |
status: | Fix Committed → Fix Released |
tags: | removed: on-verification |
According to HAProxy docs (https:/ /cbonte. github. io/haproxy- dconv/configura tion-1. 5.html# option% 20httpclose): /cbonte. github. io/haproxy- dconv/configura tion-1. 5.html# option% 20forceclose
Using "option http-server-close" or "option forceclose" is strongly recommended instead. So, I suggest to use "option forceclose":
https:/