feat: profilecli query-blocks series #3610

alsoba13 · 2024-10-07T11:17:53Z

In this PR we introduce a new profilecli command: query-blocks. With this command, you can execute queries directly to single or multiple blocks hosted in your localhost or a remote bucket.

Partially solves #3559, profile query-blocks merge will come in a later PR.

Capabilities

This feature gives similar capabilities as profilecli query seriesbut for a specified local/remote blocks:

It has support for checking label values with --label-names or specify the query with --query.
You may choose to use it locally (--local-path) or remotely (--bucket-name, --tenant-id and --object-store-type - only gcs supported right now).
You specify queried blocks with the --block-ids flag.
Time ranges (to and from) are not needed: it will query the whole blocks instead.

doc

profilecl query-blocks series --help
usage: profilecli query-blocks series [<flags>]

Request series labels on a local/remote block

Flags:
  -h, --help                     Show context-sensitive help (also try --help-long and --help-man).
      --version                  Show application version.
  -v, --verbose                  Enable verbose logging.
      --local-path="./data/anonymous/local"
                                 Path to blocks directory.
      --bucket-name=BUCKET-NAME  The name of the object storage bucket.
      --object-store-type="gcs"  The type of the object storage (e.g., gcs).
      --block-ids=BLOCK-IDS ...  List of blocks ids to query on
      --tenant-id=TENANT-ID      Tenant id of the queried block for remote bucket
      --query="{}"               Label selector to query.
      --label-names=LABEL-NAMES ...
                                 Filter returned labels to the supplied label names. Without any filter all labels are returned.

Usage example:

Querying series on a local block, and filter on service_name

profilecli query-blocks series --local-path=./data/anonymous/local --block-ids=01J9K65Z1RP3YJG66GSR9FVDQ0 --query='{service_name="ride-sharing-app"}'
level=info msg="query-block series" labelNames=[] blockIds=[01J9K65Z1RP3YJG66GSR9FVDQ0] localPath=./data/anonymous/local bucketName= tenantId=
{"__name__":"memory","__period_type__":"space","__period_unit__":"bytes","__profile_type__":"memory:alloc_objects:count:space:bytes","__service_name__":"ride-sharing-app","__type__":"alloc_objects","__unit__":"count","hostname":"6faf2cfcb5e8","pyroscope_spy":"gospy","region":"us-east","service_git_ref":"HEAD","service_name":"ride-sharing-app","service_repository":"https://github.com/grafana/pyroscope","service_root_path":"examples/language-sdk-instrumentation/golang-push/rideshare"}
{"__name__":"memory","__period_type__":"space","__period_unit__":"bytes","__profile_type__":"memory:alloc_objects:count:space:bytes","__service_name__":"ride-sharing-app","__type__":"alloc_objects","__unit__":"count","hostname":"bf91795602f5","pyroscope_spy":"gospy","region":"eu-north","service_git_ref":"HEAD","service_name":"ride-sharing-app","service_repository":"https://github.com/grafana/pyroscope","service_root_path":"examples/language-sdk-instrumentation/golang-push/rideshare"}
...

Querying series on two remote blocks, just checking at __profile_type__ label:

profilecli query-blocks series --bucket-name='dev-us-central-0-profiles-dev-001-data' --tenant-id=1218 --block-ids=01J9NBD1V2BJQWCMDQGMNGC4W5 --block-ids=01J9NBD1THECVDYVWFJVYWWX7M --label-names=__profile_type__
level=info msg="query-block series" labelNames=[__profile_type__] blockIds="[01J9NBD1V2BJQWCMDQGMNGC4W5 01J9NBD1THECVDYVWFJVYWWX7M]" localPath=./data/anonymous/local bucketName=dev-us-central-0-profiles-dev-001-data tenantId=1218
{"__profile_type__":"block:contentions:count:contentions:count"}
{"__profile_type__":"block:delay:nanoseconds:contentions:count"}
{"__profile_type__":"goroutine:goroutine:count:goroutine:count"}
{"__profile_type__":"goroutines:goroutine:count:goroutine:count"}
{"__profile_type__":"memory:alloc_objects:count:space:bytes"}
{"__profile_type__":"memory:alloc_space:bytes:space:bytes"}
{"__profile_type__":"memory:inuse_objects:count:space:bytes"}
{"__profile_type__":"memory:inuse_space:bytes:space:bytes"}
{"__profile_type__":"mutex:contentions:count:contentions:count"}
{"__profile_type__":"mutex:delay:nanoseconds:contentions:count"}
{"__profile_type__":"process_cpu:cpu:nanoseconds:cpu:nanoseconds"}
{"__profile_type__":"process_cpu:samples:count::milliseconds"}
{"__profile_type__":"process_cpu:samples:count:cpu:nanoseconds"}

alsoba13 · 2024-10-07T11:27:23Z

cmd/profilecli/output.go

extracting some common logic from query.go

alsoba13 · 2024-10-07T11:33:54Z

pkg/phlaredb/block_querier.go

@@ -159,6 +159,23 @@ func (b *BlockQuerier) BlockMetas(ctx context.Context) (metas []*block.Meta, _ e
 	return metas[0 : pos+1], nil
 }

+func (b *BlockQuerier) BlockMeta(ctx context.Context, name string) (meta *block.Meta, _ error) {


I had to introduce a direct way to read a single meta, otherwise, using BlockMetas would load every block in the bucket (potentially thousands of them)

aleks-p

Looks great to me!

Could you expand the PR description with examples of successful calls? One with local and one with remote blocks, to make it more clear on how this can be used.

kolesnikovae

Great work! Thanks for implementing this @alsoba13!

kolesnikovae · 2024-10-08T04:14:34Z

cmd/profilecli/output.go

+		for k := range m {
+			delete(m, k)
+		}


Suggested change

for k := range m {

delete(m, k)

}

clear(m)

kolesnikovae · 2024-10-08T04:18:02Z

cmd/profilecli/main.go

+	queryBlocksCmd := app.Command("query-blocks", "Query on local/remote blocks")
+	queryBlocksSeriesCmd := queryBlocksCmd.Command("series", "Request series labels on local/remote blocks")
+	queryBlocksSeriesParams := addQueryBlocksSeriesParams(queryBlocksSeriesCmd)
+


WDYT about moving this to admin blocks sub-commands?

profilecli admin blocks --help usage: profilecli admin blocks [<flags>] <command> [<args> ...] Operate on Grafana Pyroscope's blocks. Flags: -h, --help Show context-sensitive help (also try --help-long and --help-man). --version Show application version. -v, --verbose Enable verbose logging. --path="./data/local" Path to blocks directory Subcommands: admin blocks list [<flags>] List blocks. admin blocks compact [<flags>] <from> <dest> Compact blocks.

Not directly related to this PR, but if you think it makes sense we can improve the storage config handling (in all subcommands)

I think that we should use the standard client config (that supports all backends). You can register the flags as follows:

blocksCmd := adminCmd.Command("blocks", "Operate on Grafana Pyroscope's blocks.") blocksCmd.Flag("path", "Path to blocks directory").Default(defaultLocalBlockPath).StringVar(&cfg.blocks.path) var f flag.FlagSet var c client.StorageBackendConfig c.RegisterFlags(&f, logger) f.VisitAll(func(flag *flag.Flag) { blocksCmd.Flag(flag.Name, flag.Usage).Hidden().Default(flag.DefValue).SetValue(flag.Value) })

This will probably require handling of the default path and fs backend. This way we could create our Bucket object for all blocks subcommands automatically.

Regarding the flags of the storage, I'll take a look and maybe implement separately.

About making query a subcommand of admin blocks, I discarded that idea at first because I thought that profilecli admin blocks needed a pyroscope server to work. Now I see it works directly on blocks, and makes sense to merge commands here. I may implement this as well in a following PR.

marcsanmi

LGTM, just a couple minor things.

cmd/profilecli/query-blocks.go

marcsanmi · 2024-10-08T07:08:12Z

pkg/phlaredb/block_querier_test.go

+	ctx := context.Background()
+	blockQuerier := NewBlockQuerier(ctx, bucket)
+	metas, err := blockQuerier.BlockMetas(ctx)
+	require.NoError(t, err)


I'd add something like this to ensure there's actually data to test.

Suggested change

require.NoError(t, err)

require.NoError(t, err)

require.NotEmpty(t, metas)

good one as well, fixed

alsoba13 commented Oct 7, 2024

View reviewed changes

cmd/profilecli/output.go

Copy link

Contributor Author

alsoba13 Oct 7, 2024

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

extracting some common logic from query.go

alsoba13 force-pushed the alsoba13/query-blocks branch from 2e85454 to 6f7e9c5 Compare October 7, 2024 11:32

alsoba13 commented Oct 7, 2024

View reviewed changes

alsoba13 force-pushed the alsoba13/query-blocks branch from 6f7e9c5 to 7bc0d15 Compare October 7, 2024 14:32

alsoba13 marked this pull request as ready for review October 7, 2024 14:32

alsoba13 requested a review from a team as a code owner October 7, 2024 14:32

alsoba13 force-pushed the alsoba13/query-blocks branch 3 times, most recently from 3b727d3 to 40b3777 Compare October 7, 2024 14:37

feat: profilecli query-blocks series

ba8022f

alsoba13 force-pushed the alsoba13/query-blocks branch from 40b3777 to ba8022f Compare October 7, 2024 14:40

aleks-p approved these changes Oct 7, 2024

View reviewed changes

kolesnikovae approved these changes Oct 8, 2024

View reviewed changes

marcsanmi reviewed Oct 8, 2024

View reviewed changes

marcsanmi approved these changes Oct 8, 2024

View reviewed changes

alsoba13 force-pushed the alsoba13/query-blocks branch from 7035d60 to beff499 Compare October 8, 2024 09:28

pr comments

9f18c3a

alsoba13 force-pushed the alsoba13/query-blocks branch from beff499 to 9f18c3a Compare October 8, 2024 09:31

alsoba13 merged commit 426515a into main Oct 8, 2024
18 checks passed

alsoba13 deleted the alsoba13/query-blocks branch October 8, 2024 10:00

kolesnikovae mentioned this pull request Oct 10, 2024

feat: profilecli query-blocks merge #3618

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: profilecli query-blocks series #3610

feat: profilecli query-blocks series #3610

alsoba13 commented Oct 7, 2024 •

edited

Loading

alsoba13 Oct 7, 2024

alsoba13 Oct 7, 2024 •

edited

Loading

aleks-p left a comment

kolesnikovae left a comment

kolesnikovae Oct 8, 2024

kolesnikovae Oct 8, 2024

kolesnikovae Oct 8, 2024

alsoba13 Oct 8, 2024 •

edited

Loading

marcsanmi left a comment •

edited

Loading

marcsanmi Oct 8, 2024

alsoba13 Oct 8, 2024

	require.NoError(t, err)
	require.NoError(t, err)
	require.NotEmpty(t, metas)

feat: profilecli query-blocks series #3610

feat: profilecli query-blocks series #3610

Conversation

alsoba13 commented Oct 7, 2024 • edited Loading

Capabilities

doc

Usage example:

alsoba13 Oct 7, 2024

Choose a reason for hiding this comment

alsoba13 Oct 7, 2024 • edited Loading

Choose a reason for hiding this comment

aleks-p left a comment

Choose a reason for hiding this comment

kolesnikovae left a comment

Choose a reason for hiding this comment

kolesnikovae Oct 8, 2024

Choose a reason for hiding this comment

kolesnikovae Oct 8, 2024

Choose a reason for hiding this comment

kolesnikovae Oct 8, 2024

Choose a reason for hiding this comment

alsoba13 Oct 8, 2024 • edited Loading

Choose a reason for hiding this comment

marcsanmi left a comment • edited Loading

Choose a reason for hiding this comment

marcsanmi Oct 8, 2024

Choose a reason for hiding this comment

alsoba13 Oct 8, 2024

Choose a reason for hiding this comment

alsoba13 commented Oct 7, 2024 •

edited

Loading

alsoba13 Oct 7, 2024 •

edited

Loading

alsoba13 Oct 8, 2024 •

edited

Loading

marcsanmi left a comment •

edited

Loading