*: enlarge the default region-size.#17311
Conversation
This pr make the default setting of `coprocessor.region-split-size` larger, from `96MB` to `256MB` by default. The new setting for `region-split-size` is compatible to the larger cluster. Signed-off-by: lucasliang <nkcs_lykx@hotmail.com>
|
@LykxSassinator Please paste some performance comparison results here. |
Co-authored-by: Bisheng Huang <hbisheng@gmail.com> Signed-off-by: lucasliang <nkcs_lykx@hotmail.com>
Signed-off-by: lucasliang <nkcs_lykx@hotmail.com>
[LGTM Timeline notifier]Timeline:
|
|
/approve |
|
[APPROVALNOTIFIER] This PR is APPROVED This pull-request has been approved by: v01dstar, zhangjinpeng87 The full list of commands accepted by this bot can be found here. The pull request process is described here DetailsNeeds approval from an approver in each of these files:
Approvers can indicate their approval by writing |
| ## When Backup region [a,e) size exceeds `sst-max-size`, it will be backuped into several Files [a,b), | ||
| ## [b,c), [c,d), [d,e) and the size of [a,b), [b,c), [c,d) will be `sst-max-size` (or a | ||
| ## little larger). | ||
| # sst-max-size = "144MB" |
There was a problem hiding this comment.
Maybe it should be changed to the default sst file size (64MB)?
There was a problem hiding this comment.
The default value of sst_max_size is equal to region_max_size by design where it's expected to make the size of region compatible to the split size when backup.
Line 2898 in 7c50908
There was a problem hiding this comment.
In most cases, the SST size should be equal to the region size. That is we want to align the SST's boundary with the regions' because we use the SST's boundary to split regions during restoring.
This sst_max_size is used for handing abnormal sized regions -- we generate SSTs in memory, when there are huge SSTs, OOM may happen.
I guess both increase it to 384MiB or keep it unchanged are acceptable.
There was a problem hiding this comment.
The region size is uncompressed data size, but the sst size is compressed data size.
There was a problem hiding this comment.
please remember ping lightning developers when region size is changed. Some E2E failed recently.
There was a problem hiding this comment.
IMO, it's better to mark this information as an relevant info to TiKV, as the developers of TiKV may not have received this information.
|
/test |
|
@LykxSassinator: The
Use DetailsIn response to this:
Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository. |
|
/test pull-unit-test |
|
/retest |
|
@LykxSassinator: Your PR was out of date, I have automatically updated it for you. If the CI test fails, you just re-trigger the test that failed and the bot will merge the PR for you after the CI passes. DetailsInstructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the ti-community-infra/tichi repository. |
ref tikv#17309 This pr make the default setting of `coprocessor.region-split-size` larger, from `96MB` to `256MB` by default. The new setting for `region-split-size` is compatible to the larger cluster. Signed-off-by: lucasliang <nkcs_lykx@hotmail.com> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> Co-authored-by: Bisheng Huang <hbisheng@gmail.com> Signed-off-by: 3AceShowHand <jinl1037@hotmail.com>
This reverts commit 74a760f.
This reverts commit 74a760f. Signed-off-by: lucasliang <nkcs_lykx@hotmail.com>
ref pingcap/tidb#55374 Revert the changes on the configuration of region-size. These changes will be delayed until v8.4 Signed-off-by: lucasliang <nkcs_lykx@hotmail.com> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com>
This reverts commit 74a760f. Signed-off-by: lucasliang <nkcs_lykx@hotmail.com>
ref tikv#17309 This pr make the default setting of `coprocessor.region-split-size` larger, from `96MB` to `256MB` by default. The new setting for `region-split-size` is compatible to the larger cluster. Signed-off-by: lucasliang <nkcs_lykx@hotmail.com> Co-authored-by: ti-chi-bot[bot] <108142056+ti-chi-bot[bot]@users.noreply.github.com> Co-authored-by: Bisheng Huang <hbisheng@gmail.com> (cherry picked from commit 74a760f)
What is changed and how it works?
Issue Number: ref #17309
What's Changed:
Related changes
pingcap/docs/pingcap/docs-cn:Check List
Tests
Side effects
The following charts show the positive feedbacks tested on typical workloads by larger region-size (256MB), compared with the current value (96MB):
For

TPCC20k warehouses workloads (PS: As for QPS, higher is better; As for latency diff, smaller is better)For

OLTPworkload (PS: As for QPS, higher is better. And if the percent is < 0, it means "larger region-size" might give regressions. Since the fluctuations of throughput is < +-1%, it can be regarded as bias.)Release note