Connection with slave client lost.

Artem Tomyuk

unread,

Sep 26, 2016, 12:14:27 PM9/26/16

to Redis DB

Hi all.

I have a issue with diskless replication.

How it can be that connection with slave brakes just after the message that "Synchronization with slave succeeded"?

What config parameters i should increase?

So far in my conf i've set:

client-output-buffer-limit slave 0 0 0

repl-diskless-sync yes

repl-timeout 600

timeout 60

repl-backlog-size 512 mb

Log sample:

12334:M 26 Sep 18:58:44.334 * Full resync requested by slave 172.31.7.15:63221

12334:M 26 Sep 18:58:44.334 * Waiting for next BGSAVE for SYNC

15273:C 26 Sep 18:59:34.227 * DB saved on disk

15273:C 26 Sep 18:59:34.429 * RDB: 1139 MB of memory used by copy-on-write

12334:M 26 Sep 18:59:34.805 * Background saving terminated with success

12334:M 26 Sep 18:59:34.805 * Starting BGSAVE for SYNC with target: slaves sockets

12334:M 26 Sep 18:59:35.061 * Background RDB transfer started by pid 15290

15290:C 26 Sep 19:00:36.809 * RDB: 207 MB of memory used by copy-on-write

12334:M 26 Sep 19:00:37.169 * Background RDB transfer terminated with success

12334:M 26 Sep 19:00:37.169 # Slave 172.31.7.15:63221 correctly received the streamed RDB file.

12334:M 26 Sep 19:00:37.169 * Streamed RDB transfer with slave 172.31.7.15:63221 succeeded (socket). Waiting for REPLCONF ACK from slave to enable streaming

12334:M 26 Sep 19:03:25.700 * Synchronization with slave 172.31.7.15:63221 succeeded

12334:M 26 Sep 19:03:27.186 # Connection with slave client id #1244939 lost.

12334:M 26 Sep 19:03:27.702 * Slave 172.31.7.15:63221 asks for synchronization

Artem Tomyuk

unread,

Sep 26, 2016, 12:24:41 PM9/26/16

to Redis DB

on the slave side logs are:

31331:S 26 Sep 16:17:18.400 # Server started, Redis version 3.0.7

31331:S 26 Sep 16:17:18.401 * The server is now ready to accept connections on port 63221

31331:S 26 Sep 16:17:18.401 * Connecting to MASTER rdb2.leboutique.colo:63220

31331:S 26 Sep 16:17:18.403 * MASTER <-> SLAVE sync started

31331:S 26 Sep 16:17:18.403 * Non blocking connect for SYNC fired the event.

31331:S 26 Sep 16:17:18.403 * Master replied to PING, replication can continue...

31331:S 26 Sep 16:17:18.404 * Partial resynchronization not possible (no cached master)

31331:S 26 Sep 16:18:00.750 * Full resync from master: 53172658fa297f1668a191fdc2eda63dd686bcb8:17988359472

31331:S 26 Sep 16:18:01.002 * MASTER <-> SLAVE sync: receiving streamed RDB from master

31331:S 26 Sep 16:19:01.103 * MASTER <-> SLAVE sync: Flushing old data

31331:S 26 Sep 16:19:01.103 * MASTER <-> SLAVE sync: Loading DB in memory

31331:S 26 Sep 16:21:48.787 * MASTER <-> SLAVE sync: Finished with success

31331:S 26 Sep 16:21:51.170 # Connection with master lost.

31331:S 26 Sep 16:21:51.170 * Caching the disconnected master state.

31331:S 26 Sep 16:21:51.170 * Discarding previously cached master state.

понедельник, 26 сентября 2016 г., 19:14:27 UTC+3 пользователь Artem Tomyuk написал:

Sairam M P

unread,

Sep 6, 2019, 8:40:38 AM9/6/19

to Redis DB

Just noticed this thread while searching for something else. You should increase the ---------"client-output-buffer-limit slave 0 0 0"------------. Though the thread is a old one, it could help someone later.

-Sairam.

satyendra singh

unread,

Sep 9, 2019, 5:08:51 AM9/9/19

to Redis DB

Hi Sairam,

Does redisless uses disk on slave to write rdb file and load on disk?

Reply all

Reply to author

Forward