Details
-
Bug
-
Status: Done
-
Medium
-
Resolution: Fixed
-
2.29.0
-
Yes
-
6
-
Yes
-
Yes
-
No
-
Observability
-
Description
User impact:
Confusing data in Memory Available graph
Steps to reproduce:
Set public address in pmm settings
Create a pxc or psmdb cluster via dbaas
Go to mysql/mongodb cluster summary dashboard and find Node Summary stats
Set refresh rate to 5s to reproduce this quickly and observe Memory Available graph for a while
Actual result:
When we have no data from first query, we show data from the second and the graph displays two different metrcis as a result
clamp_max(avg by () (container_memory_usage_bytes{node_name=~“$node_name”,id=~“/kubepods.*“,container!~“POD|pmm-client|backup-agent|“} * 100 / container_spec_memory_limit_bytes{node_name=“$node_name”,id=~“/kubepods.*“,container!~“POD|pmm-client|backup-agent|“}),100) or
avg by () (((node_memory_MemAvailable_bytes{node_name=~“$node_name”} or (node_memory_MemFree_bytes{node_name=~“$node_name”} + node_memory_Buffers_bytes{node_name=~“$node_name”} + node_memory_Cached_bytes{node_name=~“$node_name”})) / node_memory_MemTotal_bytes{node_name=~“$node_name”}) * 100 or (100 - azure_memory_percent_average{node_name=~“$node_name”}))
Expected result:
One correct metric dispalyed
Workaround:
Details:
Attachments
Issue Links
- relates to
-
-
- Done
-