From 85aa0b2009b023041b724b1b3fc803ecd3975bbd Mon Sep 17 00:00:00 2001 From: "github-actions[bot]" Date: Fri, 12 Jun 2026 06:54:32 +0000 Subject: [PATCH] HDDS-15444. [Auto] Update configuration documentation from ozone 42eedc2529459b05928c6c287dcac4bd86d554f5 --- .../02-configuration/99-appendix.md | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/docs/05-administrator-guide/02-configuration/99-appendix.md b/docs/05-administrator-guide/02-configuration/99-appendix.md index 5b16eb73b4..2c13ef98e4 100644 --- a/docs/05-administrator-guide/02-configuration/99-appendix.md +++ b/docs/05-administrator-guide/02-configuration/99-appendix.md @@ -211,20 +211,20 @@ This page provides a comprehensive overview of the configuration keys available | `hdds.prometheus.endpoint.token` | | `SECURITY`, `MANAGEMENT` | Allowed authorization token while using prometheus servlet endpoint. This will disable SPNEGO based authentication on the endpoint. | | `hdds.public.key.file.name` | public.pem | `X509`, `SECURITY` | Name of file which stores public key generated for SCM CA. | | `hdds.raft.server.rpc.first-election.timeout` | | `OZONE`, `RATIS`, `MANAGEMENT` | ratis Minimum timeout for the first election of a leader. If not configured, fallback to `hdds.ratis.leader.election.minimum.timeout.duration`. | -| `hdds.ratis.client.exponential.backoff.base.sleep` | 4s | `OZONE`, `CLIENT`, `PERFORMANCE` | Specifies base sleep for exponential backoff retry policy. With the default base sleep of 4s, the sleep duration for ith retry is min(4 * pow(2, i), max_sleep) * r, where r is random number in the range [0.5, 1.5). | -| `hdds.ratis.client.exponential.backoff.max.retries` | 2147483647 | `OZONE`, `CLIENT`, `PERFORMANCE` | Client's max retry value for the exponential backoff policy. | -| `hdds.ratis.client.exponential.backoff.max.sleep` | 40s | `OZONE`, `CLIENT`, `PERFORMANCE` | The sleep duration obtained from exponential backoff policy is limited by the configured max sleep. Refer dfs.ratis.client.exponential.backoff.base.sleep for further details. | -| `hdds.ratis.client.multilinear.random.retry.policy` | 5s, 5, 10s, 5, 15s, 5, 20s, 5, 25s, 5, 60s, 10 | `OZONE`, `CLIENT`, `PERFORMANCE` | Specifies multilinear random retry policy to be used by ratis client. e.g. given pairs of number of retries and sleep time (n0, t0), (n1, t1), ..., for the first n0 retries sleep duration is t0 on average, the following n1 retries sleep duration is t1 on average, and so on. | -| `hdds.ratis.client.request.watch.timeout` | 3m | `OZONE`, `CLIENT`, `PERFORMANCE` | Timeout for ratis client watch request. | +| `hdds.ratis.client.exponential.backoff.base.sleep` | 1s | `OZONE`, `CLIENT`, `PERFORMANCE` | Specifies base sleep for exponential backoff retry policy. With the default base sleep of 1s, the sleep duration for ith retry is min(1 * pow(2, i), max_sleep) * r, where r is random number in the range [0.5, 1.5). | +| `hdds.ratis.client.exponential.backoff.max.retries` | 2 | `OZONE`, `CLIENT`, `PERFORMANCE` | Client's max retry value for the exponential backoff policy. | +| `hdds.ratis.client.exponential.backoff.max.sleep` | 5s | `OZONE`, `CLIENT`, `PERFORMANCE` | The sleep duration obtained from exponential backoff policy is limited by the configured max sleep. Refer dfs.ratis.client.exponential.backoff.base.sleep for further details. | +| `hdds.ratis.client.multilinear.random.retry.policy` | 5s, 6 | `OZONE`, `CLIENT`, `PERFORMANCE` | Specifies multilinear random retry policy to be used by ratis client. e.g. given pairs of number of retries and sleep time (n0, t0), (n1, t1), ..., for the first n0 retries sleep duration is t0 on average, the following n1 retries sleep duration is t1 on average, and so on. | +| `hdds.ratis.client.request.watch.timeout` | 30s | `OZONE`, `CLIENT`, `PERFORMANCE` | Timeout for ratis client watch request. | | `hdds.ratis.client.request.watch.type` | ALL_COMMITTED | `OZONE`, `CLIENT`, `PERFORMANCE` | Desired replication level when Ozone client's Raft client calls watch(), ALL_COMMITTED or MAJORITY_COMMITTED. MAJORITY_COMMITTED increases write performance by reducing watch() latency when an Ozone datanode is slow in a pipeline, at the cost of potential read latency increasing due to read retries to different datanodes. | -| `hdds.ratis.client.request.write.timeout` | 5m | `OZONE`, `CLIENT`, `PERFORMANCE` | Timeout for ratis client write request. | +| `hdds.ratis.client.request.write.timeout` | 70s | `OZONE`, `CLIENT`, `PERFORMANCE` | Timeout for ratis client write request. | | `hdds.ratis.client.retry.policy` | org.apache.hadoop.hdds.ratis.retrypolicy.RequestTypeDependentRetryPolicyCreator | `OZONE`, `CLIENT`, `PERFORMANCE` | The class name of the policy for retry. | | `hdds.ratis.client.retrylimited.max.retries` | 180 | `OZONE`, `CLIENT`, `PERFORMANCE` | Number of retries for ratis client request. | | `hdds.ratis.client.retrylimited.retry.interval` | 1s | `OZONE`, `CLIENT`, `PERFORMANCE` | Interval between successive retries for a ratis client request. | | `hdds.ratis.leader.election.minimum.timeout.duration` | 5s | `OZONE`, `RATIS`, `MANAGEMENT` | The minimum timeout duration for ratis leader election. Default is 5s. | | `hdds.ratis.raft.client.async.outstanding-requests.max` | 32 | `OZONE`, `CLIENT`, `PERFORMANCE` | Controls the maximum number of outstanding async requests that can be handled by the Standalone as well as Ratis client. | | `hdds.ratis.raft.client.rpc.request.timeout` | 60s | `OZONE`, `CLIENT`, `PERFORMANCE` | The timeout duration for ratis client request (except for watch request). It should be set greater than leader election timeout in Ratis. | -| `hdds.ratis.raft.client.rpc.watch.request.timeout` | 180s | `OZONE`, `CLIENT`, `PERFORMANCE` | The timeout duration for ratis client watch request. Timeout for the watch API in Ratis client to acknowledge a particular request getting replayed to all servers. It is highly recommended for the timeout duration to be strictly longer than Ratis server watch timeout (`hdds.ratis.raft.server.watch.timeout`) | +| `hdds.ratis.raft.client.rpc.watch.request.timeout` | 30s | `OZONE`, `CLIENT`, `PERFORMANCE` | The timeout duration for ratis client watch request. Timeout for the watch API in Ratis client to acknowledge a particular request getting replayed to all servers. It is recommended for the timeout duration to be at least as long as Ratis server watch timeout (`hdds.ratis.raft.server.watch.timeout`) | | `hdds.ratis.raft.grpc.flow.control.window` | 5MB | `OZONE`, `CLIENT`, `PERFORMANCE` | This parameter tells how much data grpc client can send to grpc server with out receiving any ack(WINDOW_UPDATE) packet from server. This parameter should be set in accordance with chunk size. Example: If Chunk size is 4MB, considering some header size in to consideration, this can be set 5MB or greater. Tune this parameter accordingly, as when it is set with a value lesser than chunk size it degrades the ozone client performance. | | `hdds.ratis.raft.server.datastream.client.pool.size` | 10 | `OZONE`, `DATANODE`, `RATIS`, `DATASTREAM` | Maximum number of client proxy in NettyServerStreamRpc for datastream write. | | `hdds.ratis.raft.server.datastream.request.threads` | 20 | `OZONE`, `DATANODE`, `RATIS`, `DATASTREAM` | Maximum number of threads in the thread pool for datastream request. |