Fix Throughput Write query in compute dashboard

Now it matches the query pattern for Throughput Read.
Not sure why `juju_unit` was missing from the `on` grouping here;
it was present in the Throughput Read query.

This should fix a bug observed where
"found duplicate serious for the match group {domain="redacted"}
on the right hand-side of the operation".
Here, the `juju_unit` label was different.

Change-Id: I5cfe7704ac8d93c4f3ff50797ec2bae3ce6dc64b
This commit is contained in:
Samuel Allan
2024-08-16 10:59:41 +09:30
parent ba7f86f174
commit 44c2d4dcb2

View File

@@ -2290,7 +2290,7 @@
"pluginVersion": "9.2.1",
"targets": [
{
"expr": "sum by(juju_unit, domain, target_device) (rate(libvirt_domain_block_stats_write_bytes_total[$__rate_interval])) * on(domain) group_left(project_name, instance_name, uuid) libvirt_domain_info_meta{project_name=~\"$project\"}",
"expr": "sum by(juju_unit, domain, target_device) (rate(libvirt_domain_block_stats_write_bytes_total[$__rate_interval])) * on(juju_unit, domain) group_left(project_name, instance_name, uuid) libvirt_domain_info_meta{project_name=~\"$project\"}",
"interval": "",
"legendFormat": "{{project_name}} | {{instance_name}} | {{uuid}} | {{target_device}}",
"refId": "A"