tools: Added thanos bucket tool rewrite (e.g allowing block series deletions). #3421

bwplotka · 2020-11-06T19:56:58Z

Added rewrite tool:

thanos tools bucket rewrite --no-dry-run \
  --id 01DN3SK96XDAEKRB1AN30AAW6E \
  --objstore.config "
type: FILESYSTEM
config:
  directory: <local dir>
" \
  --rewrite.to-delete-config "
- matchers: \"{__name__!~\\\".*total\\\"}\"
"

Not that inefficient as for first iteration. Deleting ALL gauges from 2w 8.3mln series block while gathering changelog file took ~1m30s

Head of changelog for prod block I used:

Deleted {To="1007c280a27d8401", __name__="alerts", _id="f10e61fd-069c-49eb-8ed9-b1633c6259de", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="10.35.145.22:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-control-plane-1", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1568135206056 1568178646000}]
Deleted {To="1007c280a27d8401", __name__="alerts", _id="f10e61fd-069c-49eb-8ed9-b1633c6259de", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="10.35.145.52:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-control-plane-2", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1568135206056 1568178646000}]
Deleted {To="10378dab77f14db3", __name__="alerts", _id="d6856f61-6fee-4f3c-a563-34f3d3f86e4a", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="192.168.1.43:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-master-0.bm.int.shifti.us", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1568144193553 1568156913564}]
Deleted {To="10378dab77f14db3", __name__="alerts", _id="d6856f61-6fee-4f3c-a563-34f3d3f86e4a", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="192.168.1.45:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-master-2.bm.int.shifti.us", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1568138793744 1568153673645}]
Deleted {To="1039efb496924464", __name__="alerts", _id="dee71917-dcd6-495f-bc59-59037658d9f4", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="10.48.4.213:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-control-plane-2", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1568499368241 1568499548241}]
Deleted {To="10c3d0fe3c4349f9", __name__="alerts", _id="3fec22a4-ce8b-4380-a36d-2bce5f61223e", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="172.31.136.170:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-ip-172-31-136-170.us-west-1.compute.internal", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1568117253564 1568117313545}]
Deleted {To="10feb7ef9262777e", __name__="alerts", _id="b572f112-998b-45b9-ab2f-d80f99356723", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="10.3.158.139:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-ip-10-3-158-139.eu-west-3.compute.internal", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1567802082565 1567802622565}]
Deleted {To="1170e6974f79ad38", __name__="alerts", _id="c03103eb-1571-498d-b1fd-70587b445faa", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="10.0.138.59:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-ip-10-0-138-59.ec2.internal", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1567804893521 1567870173521}]
Deleted {To="1170e6974f79ad38", __name__="alerts", _id="c03103eb-1571-498d-b1fd-70587b445faa", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="10.0.164.93:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-ip-10-0-164-93.ec2.internal", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1567805133521 1568470936744}]
Deleted {To="11c401372d21c33f", __name__="alerts", _id="b6acd906-6fc0-4d26-b4e6-e29d412c43ae", alertname="etcdMemberCommunicationSlow", alertstate="firing", endpoint="etcd-metrics", instance="192.168.50.11:9979", job="etcd", namespace="openshift-etcd", pod="etcd-member-master-1", prometheus="openshift-monitoring/k8s", prometheus_replica="prometheus-telemeter-1", service="etcd", severity="warning"} [{1568427608246 1568789465437}]

cc @SuperQ, you might find it useful (:

Signed-off-by: Bartlomiej Plotka bwplotka@gmail.com

NOTE: Left for future PRs:

Relabelling
Symbols removal / rebuild
Merge with block.Repair.

Required for CLI for deletions in Thanos: thanos-io/thanos#3421 Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>

bwplotka · 2020-11-09T01:07:36Z

Example log:

ts=2020-11-09T01:02:57.31884196Z caller=level.go:63 level=info msg="loading bucket configuration"
ts=2020-11-09T01:02:57.822533681Z caller=level.go:63 level=info msg="downloading block" source=01DN3SK96XDAEKRB1AN30AAW6E
ts=2020-11-09T01:03:10.891061066Z caller=level.go:63 level=info msg="changelog will be available" file=/tmp/thanos-rewrite/01EPN8EF1AAER6WZ6H0KSXDAEF/change.log
ts=2020-11-09T01:03:10.901763718Z caller=level.go:63 level=info msg="starting rewrite for block" source=01DN3SK96XDAEKRB1AN30AAW6E new=01EPN8EF1AAER6WZ6H0KSXDAEF toDelete="\n- matchers: \"{__name__!~\\\".*total\\\"}\"\n"
modify [{[__name__!~".*total"] []}]
ts=2020-11-09T01:03:22.477395782Z caller=level.go:63 level=info msg="processed 10.00% of 8377876 series"
ts=2020-11-09T01:03:33.502961899Z caller=level.go:63 level=info msg="processed 20.00% of 8377876 series"
ts=2020-11-09T01:03:46.514366225Z caller=level.go:63 level=info msg="processed 30.00% of 8377876 series"
ts=2020-11-09T01:04:01.191199992Z caller=level.go:63 level=info msg="processed 40.00% of 8377876 series"
ts=2020-11-09T01:04:12.062844251Z caller=level.go:63 level=info msg="processed 50.00% of 8377876 series"
ts=2020-11-09T01:04:22.684655823Z caller=level.go:63 level=info msg="processed 60.00% of 8377876 series"
ts=2020-11-09T01:04:36.599985847Z caller=level.go:63 level=info msg="processed 70.00% of 8377876 series"
ts=2020-11-09T01:04:49.444169472Z caller=level.go:63 level=info msg="processed 80.00% of 8377876 series"
ts=2020-11-09T01:05:02.105350091Z caller=level.go:63 level=info msg="processed 90.00% of 8377876 series"
ts=2020-11-09T01:05:18.180069182Z caller=level.go:63 level=info msg="processed 100.00% of 8377876 series"
ts=2020-11-09T01:05:18.180177192Z caller=level.go:63 level=info msg="wrote new block after modifications; flushing" source=01DN3SK96XDAEKRB1AN30AAW6E new=01EPN8EF1AAER6WZ6H0KSXDAEF
ts=2020-11-09T01:05:18.437332093Z caller=level.go:63 level=info msg="uploading new block" source=01DN3SK96XDAEKRB1AN30AAW6E new=01EPN8EF1AAER6WZ6H0KSXDAEF
ts=2020-11-09T01:05:18.4957816Z caller=level.go:63 level=info msg=uploaded source=01DN3SK96XDAEKRB1AN30AAW6E new=01EPN8EF1AAER6WZ6H0KSXDAEF

bwplotka · 2020-11-09T10:46:45Z

Docs preview: https://deploy-preview-3421--thanos-io.netlify.app/tip/components/tools.md/#bucket-rewrite

chadlwilson · 2020-11-09T15:52:10Z

This looks really interesting! Excuse my likely ignorance, but I have a couple of observations/queries

is there any particular reason that motivates use of the metadata.DeletionRequest format in the --rewrite.to-delete-config.* flag design here? Since logically this is doing similar stuff that a Prometheus relabel config might be doing on people's proms, I wonder whether this format (or a subset of it) might be a suitable DSL; even if action: drop is all it supports initially? Perhaps such relabel config is rather complex to translate to the actions you need to take inside the tooling - or this is to be consistent with existing "prior art" for such block-oriented tooling, however. Just a thought 👍
how are you anticipating that users will select the blocks that require rewriting/deletion of series? What should I do, if I don't know which block IDs have the problematic time series in them? Can we suggest how that could be achieved in the docs somehow?

* Exposed DeletionIterator and CompactMetas functions. Required for CLI for deletions in Thanos: thanos-io/thanos#3421 Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> * Removed Thanos usage mentions. Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>

bwplotka · 2020-11-09T18:13:08Z

Good questions @chadlwilson chadlwilson

is there any particular reason that motivates use of the metadata.DeletionRequest format in the --rewrite.to-delete-config.* flag design here? Since logically this is doing similar stuff that a Prometheus relabel config might be doing on people's proms, I wonder whether this format (or a subset of it) might be a suitable DSL; even if action: drop is all it supports initially? Perhaps such relabel config is rather complex to translate to the actions you need to take inside the tooling - or this is to be consistent with existing "prior art" for such block-oriented tooling, however. Just a thought +1

Good point. Relabelling is amazing, although a bit hard to understand and configure correctly by users. Still, it's a plan to add relabelling on top of the existing deletion request. The main problem for me is that relabelling was made for labels NOW (discovery, scraping), so it's has no time awareness. So it's hard currently to define some relabelling for a certain time range (e.g delete series only from 2:00pm to 3:00pm).

Unless we are ok to do it for the whole block always. OR we can extend relabelling to be time interval based. Any preferences?

how are you anticipating that users will select the blocks that require rewriting/deletion of series? What should I do, if I don't know which block IDs have the problematic time series in them? Can we suggest how that could be achieved in the docs somehow?

Let's add that in later PRs. Usually, you need to remove some series if you have some corruption/problem with a particular block. To print all IDs for certain labels thanos tools bucket ls or thanos tools bucket inspect are useful. On top of that thanos tools bucket web is amazing UI to view things cc @kunal-kushwaha @prmsrswt (: But good point about documenting this.

yeya24 · 2020-11-09T19:33:46Z

how are you anticipating that users will select the blocks that require rewriting/deletion of series? What should I do, if I don't know which block IDs have the problematic time series in them? Can we suggest how that could be achieved in the docs somehow?

One useful feature in promtool is the promtool tsdb analyze command, which analyzes a single TSDB block and it shows information like highest cardinality labels. It would be good if we can support this for blocks in the object storage.

But if we can expose this information via UI, then that would be perfect

pkg/compactv2/compactor.go

pkg/block/metadata/meta.go

pkg/compactv2/compactor.go

cmd/thanos/tools_bucket.go

pkg/objstore/filesystem/filesystem.go

pkg/compactv2/modifiers.go

yeya24 · 2020-11-10T03:41:40Z

If we specify a block that doesn't have any matching series of the specified deletion requests, then we will still write a new block and flush it to the obj store?
This is not a problem since we support dry-run. Just curious about whether we can improve the logic to not create a duplicate block

bwplotka · 2020-11-12T17:02:51Z

Fixed all, PTAL (:

yeya24

LGTM! It is really amazing work but it would be good to have maintainers to take a look as well since the change is huge. Maybe @kakkoyun @brancz @pracucci ?

bwplotka · 2020-11-13T10:53:24Z

PTAL 🤗

bwplotka · 2020-11-16T16:37:55Z

ping (:

docs/components/tools.md

pkg/block/index.go

pkg/block/metadata/meta.go

cmd/thanos/tools_bucket.go

pkg/block/writer.go

pkg/compactv2/compactor.go

kakkoyun · 2020-11-19T13:27:43Z

pkg/compactv2/compactor.go

+
+	symbols, set, err := compactSeries(ctx, sReaders...)
+	if err != nil {
+		return errors.Wrapf(err, "compact series from %v", func() string {


pkg/compactv2/compactor.go

bwplotka · 2020-11-23T18:47:01Z

Is this rdy to be merged? 🤗

…es deletions). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>

kakkoyun · 2020-11-24T12:57:52Z

@bwplotka Are failing tests flaky ones? In any case, I have re-run them. We'll see soon enough

kakkoyun

LGTM 💯 The flaky document checks fixed on master.

…es deletions). (thanos-io#3421) Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com> Signed-off-by: Oghenebrume50 <raphlbrume@gmail.com>

bwplotka force-pushed the bucket-delete-series branch 4 times, most recently from f82dfad to 2ad9593 Compare November 8, 2020 22:04

bwplotka added a commit to prometheus/prometheus that referenced this pull request Nov 8, 2020

Exposed DeletionIterator and CompactMetas functions.

6ba2886

Required for CLI for deletions in Thanos: thanos-io/thanos#3421 Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>

bwplotka mentioned this pull request Nov 8, 2020

Exposed DeletionIterator and CompactMetas functions. prometheus/prometheus#8161

Merged

bwplotka force-pushed the bucket-delete-series branch 4 times, most recently from 402131e to b7aefd0 Compare November 9, 2020 00:37

bwplotka marked this pull request as ready for review November 9, 2020 00:40

bwplotka force-pushed the bucket-delete-series branch from b7aefd0 to 44f00ae Compare November 9, 2020 00:45

bwplotka force-pushed the bucket-delete-series branch 4 times, most recently from de03c9b to 40ca72d Compare November 9, 2020 10:44

bwplotka requested review from yeya24, pracucci, kakkoyun and squat November 9, 2020 10:46

bwplotka force-pushed the bucket-delete-series branch 3 times, most recently from c34fb32 to 9104db5 Compare November 9, 2020 11:28

yeya24 reviewed Nov 10, 2020

View reviewed changes

bwplotka force-pushed the bucket-delete-series branch 4 times, most recently from 7cf2814 to 5590442 Compare November 12, 2020 17:02

bwplotka requested a review from yeya24 November 12, 2020 17:02

yeya24 approved these changes Nov 12, 2020

View reviewed changes

bwplotka force-pushed the bucket-delete-series branch 2 times, most recently from 7c740eb to a638117 Compare November 13, 2020 10:43

bwplotka changed the title ~~tools: Added thanos bucket tool rewrite (e.g allowing block series deletions and relabelling).~~ tools: Added thanos bucket tool rewrite (e.g allowing block series deletions). Nov 13, 2020

kakkoyun reviewed Nov 17, 2020

View reviewed changes

kakkoyun reviewed Nov 19, 2020

View reviewed changes

bwplotka force-pushed the bucket-delete-series branch 2 times, most recently from 61d1ce2 to 6fa2dc0 Compare November 20, 2020 17:34

bwplotka force-pushed the bucket-delete-series branch from 6fa2dc0 to 870a9ce Compare November 23, 2020 18:48

tools: Added thanos bucket tool rewrite (for now: allowing block seri…

d0f3fbe

…es deletions). Signed-off-by: Bartlomiej Plotka <bwplotka@gmail.com>

bwplotka force-pushed the bucket-delete-series branch from 870a9ce to d0f3fbe Compare November 23, 2020 18:49

kakkoyun approved these changes Nov 24, 2020

View reviewed changes

kakkoyun merged commit 7b45f70 into master Nov 24, 2020

bwplotka deleted the bucket-delete-series branch November 24, 2020 15:31

GiedriusS mentioned this pull request Dec 2, 2020

compact: Delete metrics in Object Storage #3529

Closed

gillg mentioned this pull request Mar 3, 2021

What do you want from Loki in 2021? grafana/loki#3119

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

tools: Added thanos bucket tool rewrite (e.g allowing block series deletions). #3421

tools: Added thanos bucket tool rewrite (e.g allowing block series deletions). #3421

bwplotka commented Nov 6, 2020 •

edited

Loading

bwplotka commented Nov 9, 2020

bwplotka commented Nov 9, 2020

chadlwilson commented Nov 9, 2020

bwplotka commented Nov 9, 2020

yeya24 commented Nov 9, 2020 •

edited

Loading

yeya24 commented Nov 10, 2020

bwplotka commented Nov 12, 2020

yeya24 left a comment

bwplotka commented Nov 13, 2020

bwplotka commented Nov 16, 2020

kakkoyun Nov 19, 2020

bwplotka Nov 19, 2020

bwplotka commented Nov 23, 2020

kakkoyun commented Nov 24, 2020

kakkoyun left a comment

tools: Added thanos bucket tool rewrite (e.g allowing block series deletions). #3421

tools: Added thanos bucket tool rewrite (e.g allowing block series deletions). #3421

Conversation

bwplotka commented Nov 6, 2020 • edited Loading

bwplotka commented Nov 9, 2020

bwplotka commented Nov 9, 2020

chadlwilson commented Nov 9, 2020

bwplotka commented Nov 9, 2020

yeya24 commented Nov 9, 2020 • edited Loading

yeya24 commented Nov 10, 2020

bwplotka commented Nov 12, 2020

yeya24 left a comment

Choose a reason for hiding this comment

bwplotka commented Nov 13, 2020

bwplotka commented Nov 16, 2020

kakkoyun Nov 19, 2020

Choose a reason for hiding this comment

bwplotka Nov 19, 2020

Choose a reason for hiding this comment

bwplotka commented Nov 23, 2020

kakkoyun commented Nov 24, 2020

kakkoyun left a comment

Choose a reason for hiding this comment

bwplotka commented Nov 6, 2020 •

edited

Loading

yeya24 commented Nov 9, 2020 •

edited

Loading