-
Notifications
You must be signed in to change notification settings - Fork 2.2k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Emit per workload labels for existing per table vttablet metrics #12394
Emit per workload labels for existing per table vttablet metrics #12394
Conversation
Review ChecklistHello reviewers! 👋 Please follow this checklist when reviewing this Pull Request. General
If a new flag is being introduced:
If a workflow is added or modified:
Bug fixes
Non-trivial changes
New/Existing features
Backward compatibility
|
This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]>
Signed-off-by: Eduardo J. Ortega U <[email protected]>
Signed-off-by: Eduardo J. Ortega U <[email protected]>
f12e5a3
to
76f5b4a
Compare
@vitessio/query-serving this needs review from y'all.. |
|
||
flagutil.DualFormatBoolVar(fs, ¤tConfig.EnablePerWorkloadTableMetrics, "enable-per-workload-table-metrics", defaultConfig.EnablePerWorkloadTableMetrics, "If true, query counts and query error metrics include a label that identifies the workload") | ||
flagutil.DualFormatStringVar(fs, ¤tConfig.WorkloadLabel, "workload-label", defaultConfig.WorkloadLabel, "The label looked for in query comments to identify the workload. Used in conjunction with enable-per-workload-table-metrics") |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I think we can avoid both of these flags. If the workload label is passed to vttablet that should be used otherwise ignored.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I believe the idea here is that the label itself is customizable. If your query comment has workload: web-app
, then the workload-label
flag value would be workload
. If you want to use xyzzy
as the label, you can do that too. The other option is to hardcode the label as workload
or some such pre-defined label. @ejortegau's approach is more flexible at the cost of adding another flag, so that's the trade-off.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
Correct, I would like to have the label user-customizable as different folks/companies may already be sending some for of this with different labels; or may refer to this in different ways. I will, however, move this to the vtgate
flags and obtain the workload information there, as suggested in the other comments.
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
I would like to keep the flag for enabling or disabling per workload metrics as this needs to be set up during statistic initialization to define whether there are 2 or 3 labels for the metrics involved - and this is before queries come in, which is when we could check whether the workload is passed to vttablet by vtgate or not. I have moved the other flag to vtgate instead, to allow the user to specify the key to look for in query directlives.
* Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]>
Signed-off-by: Eduardo J. Ortega U <[email protected]>
Signed-off-by: Eduardo J. Ortega U <[email protected]>
1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]>
Signed-off-by: Eduardo J. Ortega U <[email protected]>
Signed-off-by: Eduardo J. Ortega U <[email protected]>
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
LGTM. The docs should be change to document this new behavior and the new flag.
Docs already include both behavior and flag in this PR. |
…essio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]>
…essio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]>
…#67) * Emit per workload labels for existing per table vttablet metrics (vitessio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix cherry-pick * Add basic metrics to `vttablet` transaction throttler (vitessio#12418) * Add basic stats to vttablet tx throttler Signed-off-by: Tim Vaillancourt <[email protected]> * test new metrics Signed-off-by: Tim Vaillancourt <[email protected]> * reorder Signed-off-by: Tim Vaillancourt <[email protected]> * short names Signed-off-by: Tim Vaillancourt <[email protected]> * Add max rate Signed-off-by: Tim Vaillancourt <[email protected]> * Move NewGaugeFunc to under conditional Signed-off-by: Tim Vaillancourt <[email protected]> * Use env Signed-off-by: Tim Vaillancourt <[email protected]> * Remove env from TxThrottler struct Signed-off-by: Tim Vaillancourt <[email protected]> * Fix tests Signed-off-by: Tim Vaillancourt <[email protected]> * PR suggestion Signed-off-by: Tim Vaillancourt <[email protected]> * Fix unit test Signed-off-by: Tim Vaillancourt <[email protected]> * reorder test vars Signed-off-by: Tim Vaillancourt <[email protected]> --------- Signed-off-by: Tim Vaillancourt <[email protected]> * Fix go/vt/vttablet/tabletserver/query_engine_test.go --------- Signed-off-by: Eduardo J. Ortega U <[email protected]> Signed-off-by: Tim Vaillancourt <[email protected]> Co-authored-by: Eduardo J. Ortega U <[email protected]>
This PR backports the functionality of vitessio#12394 to our v12 fork. Some changes were necessary with respect to the upstream PR due to the distance between v12 and the v16+ code against which the upstream change was built - particularly around the area of query directive extraction; as well as the metrics that are instrumented. This will not need to be ported to v14 it has been already included in #67 .
…essio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]>
* Emit per workload labels for existing per table vttablet metrics (vitessio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix bad conflict resolution --------- Signed-off-by: Eduardo J. Ortega U <[email protected]> Co-authored-by: Eduardo J. Ortega U <[email protected]>
…essio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]>
…essio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]>
* Add basic metrics to `vttablet` transaction throttler (vitessio#12418) * Add basic stats to vttablet tx throttler Signed-off-by: Tim Vaillancourt <[email protected]> * test new metrics Signed-off-by: Tim Vaillancourt <[email protected]> * reorder Signed-off-by: Tim Vaillancourt <[email protected]> * short names Signed-off-by: Tim Vaillancourt <[email protected]> * Add max rate Signed-off-by: Tim Vaillancourt <[email protected]> * Move NewGaugeFunc to under conditional Signed-off-by: Tim Vaillancourt <[email protected]> * Use env Signed-off-by: Tim Vaillancourt <[email protected]> * Remove env from TxThrottler struct Signed-off-by: Tim Vaillancourt <[email protected]> * Fix tests Signed-off-by: Tim Vaillancourt <[email protected]> * PR suggestion Signed-off-by: Tim Vaillancourt <[email protected]> * Fix unit test Signed-off-by: Tim Vaillancourt <[email protected]> * reorder test vars Signed-off-by: Tim Vaillancourt <[email protected]> --------- Signed-off-by: Tim Vaillancourt <[email protected]> * Fix transaction throttler ignoring the initial rate (vitessio#12618) * Fix transaction throttler ignoring the initial rate This addresses the issue reported in vitessio#12549 Signed-off-by: Eduardo J. Ortega U <[email protected]> * Add missing override of max replication lag in `throttler.newThrottler()` Signed-off-by: Eduardo J. Ortega U <[email protected]> * Reorder functions to make diff easier to read Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix check for maxRate in `newThrottlerFromConfig()` Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix some CI pipeline issues Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix typo Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]> Signed-off-by: Eduardo J. Ortega U. <[email protected]> * Cleanup panics in `txthrottler`, reorder for readability (vitessio#12901) * Cleanup tx_throttler.go Signed-off-by: Tim Vaillancourt <[email protected]> * Cleanup tx_throttler.go #2 Signed-off-by: Tim Vaillancourt <[email protected]> * Fix throttlerFactoryFunc Signed-off-by: Tim Vaillancourt <[email protected]> * Undo if-cond consolidation Signed-off-by: Tim Vaillancourt <[email protected]> * Undo struct shuffling Signed-off-by: Tim Vaillancourt <[email protected]> * prove that disabled config returns nil error Signed-off-by: Tim Vaillancourt <[email protected]> * Improve test Signed-off-by: Tim Vaillancourt <[email protected]> --------- Signed-off-by: Tim Vaillancourt <[email protected]> * Emit per workload labels for existing per table vttablet metrics (vitessio#12394) * Emit per workload labels for existing per table vttablet metrics This adds the possibility to configure vttablet (via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else. This is usefult to be able to gain observability about how the query load is distributed across different workloads. This is achieved with two new CLI flags, namely: * `enable-per-workload-table-metrics`: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. * `workload-label`: a string to look for in query comments to identify the workload running the current query. The workload is obtained by parsing query comments of the form: /* ... <workload_label>=<workload_name>; ... */ For example, if vttablet is started with `--enable-per-workload-table-metrics --workload-label app_name` anda query is issued with a comment like /* ... app_name=shop; ... */ then metrics will look like ``` vttablet_query_counts{plan="Select",table="dual", workload="shop"} 15479 ``` instead of ``` vttablet_query_counts{plan="Select",table="dual"} 15479 ``` Query comment parsing only takes place if `--enable-per-workload-table-metrics` is used, as to not incur parsing performance impact if the user does not want per workload metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * make linter happy Signed-off-by: Eduardo J. Ortega U <[email protected]> * fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments: * Obtain workload information on the vtgate instead of the vttablet, avoiding double parsing. * Treat workload name as a query directive. * Send workload name from vtgate to vttablet as ExecuteOptions. Additionally, annotate tabletserver's execution span with the workload name to also enrich traces with workload name data, in addition to metrics. Signed-off-by: Eduardo J. Ortega U <[email protected]> * A few fixes: 1. Rebuild some files with `make proto`. 2. Protect against nil ExecuteOptions on the tabletserver. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix flags e2e test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comments Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fixes Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix a comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Update JS code for protobuf changes. Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix QueryEngine unit test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix e2e flag test Signed-off-by: Eduardo J. Ortega U <[email protected]> * Fix spurious tab in comment Signed-off-by: Eduardo J. Ortega U <[email protected]> * Address PR comment Don't use dual format flag for new flags - stick with - separated ones. Signed-off-by: Eduardo J. Ortega U <[email protected]> --------- Signed-off-by: Eduardo J. Ortega U <[email protected]> * remove mistaken git add Signed-off-by: Tim Vaillancourt <[email protected]> * make vtadmin_web_proto_types Signed-off-by: Tim Vaillancourt <[email protected]> * test unit_race test on go-version: 1.18.9 Signed-off-by: Tim Vaillancourt <[email protected]> * Revert "test unit_race test on go-version: 1.18.9" This reverts commit 922e897. * CI: Misc test improvements to limit failures with various runners (vitessio#13825) Signed-off-by: Matt Lord <[email protected]> * Fix setup order to avoid races (vitessio#13871) Signed-off-by: Dirkjan Bussink <[email protected]> --------- Signed-off-by: Tim Vaillancourt <[email protected]> Signed-off-by: Eduardo J. Ortega U <[email protected]> Signed-off-by: Eduardo J. Ortega U. <[email protected]> Signed-off-by: Matt Lord <[email protected]> Signed-off-by: Dirkjan Bussink <[email protected]> Co-authored-by: Eduardo J. Ortega U <[email protected]> Co-authored-by: Matt Lord <[email protected]> Co-authored-by: Dirkjan Bussink <[email protected]>
Description
This adds the possibility to configure
vttablet
(via CLI flag) to also have a workload label for existing per table metrics (query counts, query times, query errors, query rows affected, query rows returned, query error counts). Workload can be any string that makes sense for the client application. For example, API endpoint name, controller, batch job name, application name or something else.This is useful to be able to gain observability about how the query load is distributed across different workloads.
This is achieved with a new CLI flag, namely:
enable-per-workload-table-metrics
: whether to enable or disable per workload metric collection - disabled by default to preserve the current behavior, thus making the new feature opt-in only. This flag is forvttablet
.The workload is obtained by parsing query directive comments of the form:
For example, if
vttablet
is started with--enable-per-workload-table-metrics
, and query is issued with a comment likethen metrics will look like
instead of
The workload name is also being added to the tracing span.
Related Issue(s)
#12490
Checklist
Deployment Notes
The new behavior with respect to metrics is disabled by default, and will only be enabled if the relevant CLI flag is passed when starting the
vttablet
process.