Tarantool development patches archive
 help / color / mirror / Atom feed
From: Serge Petrenko <sergepetrenko@tarantool.org>
To: vdavydov.dev@gmail.com
Cc: tarantool-patches@freelists.org,
	Serge Petrenko <sergepetrenko@tarantool.org>
Subject: [PATCH v3] replication: enter orphan mode on manual replication configuration chage
Date: Wed, 28 Aug 2019 14:20:28 +0300	[thread overview]
Message-ID: <20190828112028.13875-1-sergepetrenko@tarantool.org> (raw)

Currently we only enter orphan mode when instance fails to connect to
replication_connect_quorum remote instances during local recovery.
On bootstrap and manual replication configuration change an error is
thrown. We better enter orphan mode on manual config change, and leave
it only in case we managed to sync with replication_connect_quorum
instances.

Closes #4424

@TarantoolBot document
Title: document reaction on error in replication configuration change.

Now when issuing `box.cfg{replication={uri1, uri2, ...}}` and failing to
sync with replication_connect_quorum remote instances, the server will
throw an error, if it is bootstrapping, and just set its state to orphan
in all other cases (recovering from existing xlog/snap files or manually
changing box.cfg.replication on the fly). To leave orphan mode, you may
wait until the server manages to sync with replication_connect_quorum
instances.
In order to leave orphan mode you need to make the server sync with
enough instances. To do so, you may either:
1) set replication_connect_quorum to a lower value
2) reset box.cfg.replication to exclude instances that cannot
   be reached or synced with
3) just set box.cfg.replication to "" (empty string)
---
https://github.com/tarantool/tarantool/issues/4424
https://github.com/tarantool/tarantool/tree/sp/gh-4424-repl-config-errors-v33

 src/box/box.cc                 |  14 +++--
 src/box/box.h                  |   7 +++
 src/box/replication.cc         |  11 ++++
 test/replication/misc.result   | 104 ++++++++++++++++++++++++++++++++-
 test/replication/misc.test.lua |  41 ++++++++++++-
 5 files changed, 168 insertions(+), 9 deletions(-)

diff --git a/src/box/box.cc b/src/box/box.cc
index 66cd6d3a4..62d9665f4 100644
--- a/src/box/box.cc
+++ b/src/box/box.cc
@@ -257,7 +257,7 @@ box_wait_ro(bool ro, double timeout)
 }
 
 void
-box_set_orphan(bool orphan)
+box_set_orphan_raw(bool orphan)
 {
 	if (is_orphan == orphan)
 		return; /* nothing to do */
@@ -266,7 +266,12 @@ box_set_orphan(bool orphan)
 
 	is_orphan = orphan;
 	fiber_cond_broadcast(&ro_cond);
+}
 
+void
+box_set_orphan(bool orphan)
+{
+	box_set_orphan_raw(orphan);
 	/* Update the title to reflect the new status. */
 	if (is_orphan) {
 		say_crit("entering orphan mode");
@@ -699,11 +704,10 @@ box_set_replication(void)
 	box_check_replication();
 	/*
 	 * Try to connect to all replicas within the timeout period.
-	 * The configuration will succeed as long as we've managed
-	 * to connect to at least replication_connect_quorum
-	 * masters.
+	 * Stay in orphan mode in case we fail to connect to at least
+	 * 'replication_connect_quorum' remote instances.
 	 */
-	box_sync_replication(true);
+	box_sync_replication(false);
 	/* Follow replica */
 	replicaset_follow();
 	/* Wait until appliers are in sync */
diff --git a/src/box/box.h b/src/box/box.h
index ddcfbe2e5..3bd484353 100644
--- a/src/box/box.h
+++ b/src/box/box.h
@@ -127,6 +127,13 @@ box_wait_ro(bool ro, double timeout);
 void
 box_set_orphan(bool orphan);
 
+/**
+ * Set orphan mode but don't update instance title.
+ * \sa box_set_orphan
+ */
+void
+box_set_orphan_raw(bool orphan);
+
 /**
  * Iterate over all spaces and save them to the
  * snapshot file.
diff --git a/src/box/replication.cc b/src/box/replication.cc
index 28f7acedc..ac415fd46 100644
--- a/src/box/replication.cc
+++ b/src/box/replication.cc
@@ -610,6 +610,17 @@ replicaset_connect(struct applier **appliers, int count,
 
 	say_info("connecting to %d replicas", count);
 
+	if (!connect_quorum) {
+		/*
+		 * Enter orphan mode on configuration change and
+		 * only leave it when we manage to sync with
+		 * replicaset_quorum instances. Don't change
+		 * title though, it should be 'loading' during
+		 * local recovery.
+		 */
+		box_set_orphan_raw(true);
+	}
+
 	/*
 	 * Simultaneously connect to remote peers to receive their UUIDs
 	 * and fill the resulting set:
diff --git a/test/replication/misc.result b/test/replication/misc.result
index 0a57edda5..ae72ce3e4 100644
--- a/test/replication/misc.result
+++ b/test/replication/misc.result
@@ -18,10 +18,19 @@ replication_connect_timeout = box.cfg.replication_connect_timeout
 box.cfg{replication_timeout=0.05, replication_connect_timeout=0.05, replication={}}
 ---
 ...
+box.cfg{replication_connect_quorum=2}
+---
+...
 box.cfg{replication = {'127.0.0.1:12345', box.cfg.listen}}
 ---
-- error: 'Incorrect value for option ''replication'': failed to connect to one or
-    more replicas'
+...
+box.info.status
+---
+- orphan
+...
+box.info.ro
+---
+- true
 ...
 -- gh-3606 - Tarantool crashes if box.cfg.replication is updated concurrently
 fiber = require('fiber')
@@ -47,8 +56,16 @@ c:get()
 ---
 - true
 ...
-box.cfg{replication_timeout = replication_timeout, replication_connect_timeout = replication_connect_timeout}
+box.cfg{replication = "", replication_timeout = replication_timeout, replication_connect_timeout = replication_connect_timeout}
+---
+...
+box.info.status
+---
+- running
+...
+box.info.ro
 ---
+- false
 ...
 -- gh-3111 - Allow to rebootstrap a replica from a read-only master
 replica_uuid = uuid.new()
@@ -729,3 +746,84 @@ test_run:cleanup_cluster()
 box.schema.user.revoke('guest', 'replication')
 ---
 ...
+--
+-- gh-4424 Always enter orphan mode on error in replication
+-- configuration change.
+--
+replication_connect_timeout = box.cfg.replication_connect_timeout
+---
+...
+replication_connect_quorum = box.cfg.replication_connect_quorum
+---
+...
+box.cfg{replication="12345", replication_connect_timeout=0.1, replication_connect_quorum=1}
+---
+...
+box.info.status
+---
+- orphan
+...
+box.info.ro
+---
+- true
+...
+-- reset replication => leave orphan mode
+box.cfg{replication=""}
+---
+...
+box.info.status
+---
+- running
+...
+box.info.ro
+---
+- false
+...
+-- no switch to orphan when quorum == 0
+box.cfg{replication="12345", replication_connect_quorum=0}
+---
+...
+box.info.status
+---
+- running
+...
+box.info.ro
+---
+- false
+...
+-- we could connect to one out of two replicas. Set orphan.
+box.cfg{replication_connect_quorum=2}
+---
+...
+box.cfg{replication={box.cfg.listen, "12345"}}
+---
+...
+box.info.status
+---
+- orphan
+...
+box.info.ro
+---
+- true
+...
+-- lower quorum => leave orphan mode
+box.cfg{replication_connect_quorum=1}
+---
+...
+box.info.status
+---
+- running
+...
+box.info.ro
+---
+- false
+...
+box.cfg{replication=""}
+---
+...
+box.cfg{replication_connect_timeout=replication_connect_timeout}
+---
+...
+box.cfg{replication_connect_quorum=replication_connect_quorum}
+---
+...
diff --git a/test/replication/misc.test.lua b/test/replication/misc.test.lua
index 99e995509..16e7e9e42 100644
--- a/test/replication/misc.test.lua
+++ b/test/replication/misc.test.lua
@@ -8,7 +8,10 @@ box.schema.user.grant('guest', 'replication')
 replication_timeout = box.cfg.replication_timeout
 replication_connect_timeout = box.cfg.replication_connect_timeout
 box.cfg{replication_timeout=0.05, replication_connect_timeout=0.05, replication={}}
+box.cfg{replication_connect_quorum=2}
 box.cfg{replication = {'127.0.0.1:12345', box.cfg.listen}}
+box.info.status
+box.info.ro
 
 -- gh-3606 - Tarantool crashes if box.cfg.replication is updated concurrently
 fiber = require('fiber')
@@ -19,7 +22,9 @@ f()
 c:get()
 c:get()
 
-box.cfg{replication_timeout = replication_timeout, replication_connect_timeout = replication_connect_timeout}
+box.cfg{replication = "", replication_timeout = replication_timeout, replication_connect_timeout = replication_connect_timeout}
+box.info.status
+box.info.ro
 
 -- gh-3111 - Allow to rebootstrap a replica from a read-only master
 replica_uuid = uuid.new()
@@ -293,3 +298,37 @@ test_run:cmd("cleanup server replica")
 test_run:cmd("delete server replica")
 test_run:cleanup_cluster()
 box.schema.user.revoke('guest', 'replication')
+
+--
+-- gh-4424 Always enter orphan mode on error in replication
+-- configuration change.
+--
+replication_connect_timeout = box.cfg.replication_connect_timeout
+replication_connect_quorum = box.cfg.replication_connect_quorum
+box.cfg{replication="12345", replication_connect_timeout=0.1, replication_connect_quorum=1}
+box.info.status
+box.info.ro
+-- reset replication => leave orphan mode
+box.cfg{replication=""}
+box.info.status
+box.info.ro
+-- no switch to orphan when quorum == 0
+box.cfg{replication="12345", replication_connect_quorum=0}
+box.info.status
+box.info.ro
+
+-- we could connect to one out of two replicas. Set orphan.
+box.cfg{replication_connect_quorum=2}
+box.cfg{replication={box.cfg.listen, "12345"}}
+box.info.status
+box.info.ro
+-- lower quorum => leave orphan mode
+box.cfg{replication_connect_quorum=1}
+box.info.status
+box.info.ro
+
+box.cfg{replication=""}
+
+
+box.cfg{replication_connect_timeout=replication_connect_timeout}
+box.cfg{replication_connect_quorum=replication_connect_quorum}
-- 
2.20.1 (Apple Git-117)

             reply	other threads:[~2019-08-28 11:20 UTC|newest]

Thread overview: 3+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2019-08-28 11:20 Serge Petrenko [this message]
2019-08-28 11:50 ` [tarantool-patches] " Konstantin Osipov
2019-08-28 13:58 ` Vladimir Davydov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20190828112028.13875-1-sergepetrenko@tarantool.org \
    --to=sergepetrenko@tarantool.org \
    --cc=tarantool-patches@freelists.org \
    --cc=vdavydov.dev@gmail.com \
    --subject='Re: [PATCH v3] replication: enter orphan mode on manual replication configuration chage' \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox