cloud_storage: Increase disk tput limit a bit #18451
Conversation
When setting read-path tput limit increase disk limit by 6%
| @@ -144,8 +144,9 @@ get_throughput_limit(std::optional<size_t> device_throughput) { | |||
| } | |||
|
|
|||
| auto tp = std::min(hard_limit, device_throughput.value()); | |||
| constexpr auto tput_overshoot_frac = 16; | |||
There was a problem hiding this comment.
Is there anything special about this number? Or was it empirically determined?
If empirically determined, how did you determine how effective this PR is (in case we need to adjust it further down the line)?
There was a problem hiding this comment.
it's a simple back of the envelope calculation (2GiB/s with 50% allowance for TS will give us around 60MiB/s per node)
we only need to limit disk as a last resort, the code is relying on network limit which is set per shard
There was a problem hiding this comment.
it's a simple back of the envelope calculation (2GiB/s with 50% allowance for TS will give us around 60MiB/s per node)
we only need to limit disk as a last resort, the code is relying on network limit which is set per shard
add to commit message?
|
/ci-repeat 10 |
|
/cdt |
1 similar comment
|
/cdt |
| @@ -144,8 +144,9 @@ get_throughput_limit(std::optional<size_t> device_throughput) { | |||
| } | |||
|
|
|||
| auto tp = std::min(hard_limit, device_throughput.value()); | |||
| constexpr auto tput_overshoot_frac = 16; | |||
There was a problem hiding this comment.
it's a simple back of the envelope calculation (2GiB/s with 50% allowance for TS will give us around 60MiB/s per node)
we only need to limit disk as a last resort, the code is relying on network limit which is set per shard
add to commit message?
|
This continues to fail #14225 cdt run from may 21 https://buildkite.com/redpanda/vtools/builds/13944#018f9a62-23b0-4336-8728-2860552ef36e |
When setting read-path tput limit increase disk limit by 6%
Fixes #14139
After the Seastar update the throttling test started to fail because disk limit is now more strict. This PR relaxes the limit a bit so the network could be throttled in the test. This should also improve observability a bit because Seastar does not expose any disk throttling metrics but we do have network throttling metrics. By having disk limit slightly higher we will be always hitting throttling in our net layer which has better metrics. There should be no impact for anything else.
Backports Required
Release Notes