You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@kudu.apache.org by ad...@apache.org on 2019/11/18 05:24:31 UTC
[kudu] branch master updated (a0e8964 -> 24cf1c9)
This is an automated email from the ASF dual-hosted git repository.
adar pushed a change to branch master
in repository https://gitbox.apache.org/repos/asf/kudu.git.
from a0e8964 [java] fixed bug in the connection negotiation code
new 200509b docs: add docs for maintenance mode
new 24cf1c9 [cfile] No longer need to `relocate()` in `ReadBlock()`
The 2 revisions listed above as "new" are entirely new to this
repository and will be described in separate emails. The revisions
listed as "add" were already present in the repository and have only
been added to this reference.
Summary of changes:
docs/administration.adoc | 45 ++++++++++++++++++++++++++++++++++--------
src/kudu/cfile/cfile_reader.cc | 6 ------
2 files changed, 37 insertions(+), 14 deletions(-)
[kudu] 01/02: docs: add docs for maintenance mode
Posted by ad...@apache.org.
This is an automated email from the ASF dual-hosted git repository.
adar pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/kudu.git
commit 200509b497a028f7d700152ca4c92818ce5b9f0e
Author: Andrew Wong <aw...@apache.org>
AuthorDate: Fri Nov 15 12:33:05 2019 -0800
docs: add docs for maintenance mode
A staged version can be found here:
https://github.com/andrwng/kudu/blob/docs_maintenance_mode/docs/administration.adoc#minimizing_cluster_disruption_during_temporary_single_ts_downtime
Change-Id: I36b9eddc1d4d4a4e4cb149058fc6d6f438e47f1f
Reviewed-on: http://gerrit.cloudera.org:8080/14718
Reviewed-by: Alexey Serbin <as...@cloudera.com>
Tested-by: Kudu Jenkins
---
docs/administration.adoc | 45 +++++++++++++++++++++++++++++++++++++--------
1 file changed, 37 insertions(+), 8 deletions(-)
diff --git a/docs/administration.adoc b/docs/administration.adoc
index 2827235..f8a6091 100644
--- a/docs/administration.adoc
+++ b/docs/administration.adoc
@@ -1461,14 +1461,43 @@ for more than `--follower_unavailable_considered_failed_sec` (default 300)
seconds, the tablet replicas on the down tablet server will be replaced by new
replicas on available tablet servers. This will cause stress on the cluster
as tablets re-replicate and, if the downtime lasts long enough, significant
-reduction in the number of replicas on the down tablet server. This may require
-the rebalancer to fix.
+reduction in the number of replicas on the down tablet server, which would
+require the rebalancer to fix.
-To work around this, increase `--follower_unavailable_considered_failed_sec` on
-all tablet servers so the amount of time before re-replication will start is
-longer than the expected downtime of the tablet server, including the time it
-takes the tablet server to restart and bootstrap its tablet replicas. To do
-this, run the following command for each tablet server:
+To work around this, in Kudu versions from 1.11 onwards, the `kudu` CLI
+contains a tool to put tablet servers into maintenance mode. While in this
+state, the tablet server’s replicas are not re-replicated due to its downtime
+alone, though re-replication may still occur in the event that the server in
+maintenance suffers from a disk failure or if a follower replica on the tablet
+server falls too far behind its leader replica. Upon exiting maintenance,
+re-replication is triggered for any remaining under-replicated tablets.
+
+The `kudu tserver state enter_maintenance` and `kudu tserver state
+exit_maintenance` tools are added to orchestrate tablet server maintenance.
+The following can be run from a tablet server to put it into maintenance:
+
+[source,bash]
+----
+$ TS_UUID=$(sudo -u kudu kudu fs dump uuid --fs_wal_dir=<wal_dir> --fs_data_dirs=<data_dirs>)
+$ sudo -u kudu kudu tserver state enter_maintenance <master_addresses> "$TS_UUID"
+----
+
+The tablet server maintenance mode is shown in the "Tablet Servers" page of the
+Kudu leader master's web UI, and in the output of `kudu cluster ksck`. To exit
+maintenance mode, run the following:
+
+[source,bash]
+----
+$ sudo -u kudu kudu tserver state exit_maintenance <master_addresses> "$TS_UUID"
+----
+
+In versions prior to 1.11, a different approach must be used to prevent
+unwanted re-replication. Increase
+`--follower_unavailable_considered_failed_sec` on all tablet servers so the
+amount of time before re-replication starts is longer than the expected
+downtime of the tablet server, including the time it takes the tablet server to
+restart and bootstrap its tablet replicas. To do this, run the following
+command for each tablet server:
[source,bash]
----
@@ -1486,7 +1515,7 @@ WARNING: Be sure to reset the value of `--follower_unavailable_considered_failed
to its original value.
NOTE: On Kudu versions prior to 1.8, the `--force` flag must be provided in the above
-commands.
+`set_flag` commands.
[[rebalancer_tool]]
=== Running the tablet rebalancing tool
[kudu] 02/02: [cfile] No longer need to `relocate()` in
`ReadBlock()`
Posted by ad...@apache.org.
This is an automated email from the ASF dual-hosted git repository.
adar pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/kudu.git
commit 24cf1c948d07676cfc567b73cef7792a38d1809e
Author: lingbin <li...@gmail.com>
AuthorDate: Tue Nov 12 11:50:57 2019 +0800
[cfile] No longer need to `relocate()` in `ReadBlock()`
In the current implementation, when reading data from
`ReadableBlock`, the `Slice` object will no longer be
modified, but just fill it with data. That is,
slice->data() will always match pre-allocated buffer.
Change-Id: I1619d05d42c9ec20e029097aaaf37b3ac93b9a69
Reviewed-on: http://gerrit.cloudera.org:8080/14693
Reviewed-by: Adar Dembo <ad...@cloudera.com>
Tested-by: Kudu Jenkins
---
src/kudu/cfile/cfile_reader.cc | 6 ------
1 file changed, 6 deletions(-)
diff --git a/src/kudu/cfile/cfile_reader.cc b/src/kudu/cfile/cfile_reader.cc
index 12794a0..5ebc193 100644
--- a/src/kudu/cfile/cfile_reader.cc
+++ b/src/kudu/cfile/cfile_reader.cc
@@ -536,12 +536,6 @@ Status CFileReader::ReadBlock(const IOContext* io_context, const BlockPointer &p
// Set the result block to our decompressed data.
block = Slice(buf, uncompressed_size);
- } else {
- // Some of the File implementations from LevelDB attempt to be tricky
- // and just return a Slice into an mmapped region (or in-memory region).
- // But, this is hard to program against in terms of cache management, etc,
- // so we memcpy into our scratch buffer if necessary.
- block.relocate(scratch.get());
}
// It's possible that one of the TryAllocateFromCache() calls above