Restore: add bandwidth metrics #4081
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
This should make it easier to see what is updated where and when.
Michal-Leszczynski
force-pushed
the
ml/restore-bw-metrics
branch
from
98faa03 to
b4f6671
Compare
They are really useful for evaluating restore performance.
It's useful for checking/tracking restore performance. Ref #4042
Michal-Leszczynski
force-pushed
the
ml/restore-bw-metrics
branch
from
b4f6671 to
2ba54e2
Compare
This was a left-over from the PR introducing indexing (14aef7b). It also initialized metrics as a part of the indexing procedure, but it forgot to remove the previous metrics initialization from the code.
There was a confusion about which cluster ID should be used for labeling remaining_bytes metric. When setting remaining_bytes, we used backup cluster ID, but when decreasing, we used restore cluster ID. Backup cluster ID should be used in both places as this metrics describes how many bytes from which place are yet to be restored. Since we use backup cluster DC, node ID, etc., we should also use backup cluster ID.
|
Examples for a single restore worker: Streamed bytes
Downloaded bytes (per location)
Stream bandwidth in MB/s
Download bandwidth in MB/s
|
|
@karol-kokoszka This PR is ready for review! |
karol-kokoszka
approved these changes
Oct 29, 2024
karol-kokoszka
pushed a commit
that referenced
this pull request
Nov 4, 2024
* refactor(restore): separate methods for updating metrics/progress This should make it easier to see what is updated where and when. * feat(metrics): restore, add bandwidth metrics They are really useful for evaluating restore performance. * feat(restore): set download/stream bytes/duration metrics It's useful for checking/tracking restore performance. Ref #4042 * fix(restore): don't initialize metrics twice This was a left-over from the PR introducing indexing (14aef7b). It also initialized metrics as a part of the indexing procedure, but it forgot to remove the previous metrics initialization from the code. * fix(restore): use backup bluster ID in remaining_bytes metric There was a confusion about which cluster ID should be used for labeling remaining_bytes metric. When setting remaining_bytes, we used backup cluster ID, but when decreasing, we used restore cluster ID. Backup cluster ID should be used in both places as this metrics describes how many bytes from which place are yet to be restored. Since we use backup cluster DC, node ID, etc., we should also use backup cluster ID.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
This PR add new restore metrics:
downloaded_bytes(downloaded bytes) labeled bycluster,location,hostdownload_duration(download duration in ms) labeled bycluster,location,hoststreamed_bytes(load&streamed bytes) labeled bycluster,hoststream_duration(load&stream duration in ms) labeled bycluster,hostThey can be easily used to calculate download/stream bandwidth.
Ref #4042