Tarantool do self bootstrap in read_only mode if failed to any replica when `replication_connect_quorum' is 0

[rtokarev]

Tarantool version: 1.10.3-106-g4faa103

OS version: CentOS 6

Bug description:

When `replication_connect_quorum' is configured to 0, tarantool do self bootstrap (creates own snapshot) even if configured read only. It violates the read only principe.

Steps to reproduce:

[root@st-drbd1 1]# tarantool
Tarantool 1.10.3-106-g4faa103
type 'help' for interactive help
tarantool> box.cfg{ read_only = true, replication = { '0:12345' }, replication_connect_quorum = 0, replication_connect_timeout = 1 }
2019-08-12 18:40:20.072 [17248] main/101/interactive C> Tarantool 1.10.3-106-g4faa103
2019-08-12 18:40:20.072 [17248] main/101/interactive C> log level 5
2019-08-12 18:40:20.072 [17248] main/101/interactive I> mapping 268435456 bytes for memtx tuple arena...
2019-08-12 18:40:20.073 [17248] main/101/interactive I> mapping 134217728 bytes for vinyl tuple arena...
2019-08-12 18:40:20.083 [17248] main/101/interactive I> instance uuid 674ba58c-ebcc-4bfb-a886-acd828b41133
2019-08-12 18:40:20.083 [17248] main/101/interactive I> connecting to 1 replicas
2019-08-12 18:40:20.084 [17248] main/105/applier/0:12345 I> can't connect to master
2019-08-12 18:40:20.084 [17248] main/105/applier/0:12345 coio.cc:106 !> SystemError connect, called on fd 11, aka 127.0.0.1:39484: Connection refused
2019-08-12 18:40:20.084 [17248] main/105/applier/0:12345 I> will retry every 1.00 second
2019-08-12 18:40:21.084 [17248] main/101/interactive C> failed to connect to 1 out of 1 replicas
2019-08-12 18:40:21.084 [17248] main/101/interactive I> initializing an empty data directory
2019-08-12 18:40:21.152 [17248] main/101/interactive I> assigned id 1 to replica 674ba58c-ebcc-4bfb-a886-acd828b41133
2019-08-12 18:40:21.152 [17248] main/101/interactive I> cluster uuid d564b319-061f-4be6-9e45-a325aa524889
2019-08-12 18:40:21.166 [17248] snapshot/101/main I> saving snapshot `./00000000000000000000.snap.inprogress'
2019-08-12 18:40:21.211 [17248] snapshot/101/main I> done
2019-08-12 18:40:21.212 [17248] main/101/interactive I> ready to accept requests
2019-08-12 18:40:21.214 [17248] main/109/checkpoint_daemon I> started
2019-08-12 18:40:21.214 [17248] main/109/checkpoint_daemon I> scheduled the next snapshot at Mon Aug 12 20:23:53 2019
2019-08-12 18:40:21.214 [17248] main/101/interactive I> set 'read_only' configuration option to true
---
...

tarantool> [root@st-drbd1 1]# ls -l
total 8
-rw-r--r-- 1 root root 1611 Aug 12 18:40 00000000000000000000.snap
-rw-r--r-- 1 root root  102 Aug 12 18:40 00000000000000000000.xlog

[kostja]

This was "fixed" by #3428
So this is apparently not a bug.

[kyukhin]

Closing then.