Hope you guys can help me out with an issue we are having. First of, our setup constitutes of all canvas components running in a single server with plenty of CPU, RAM and storage. We have just recently started experiencing some random timeouts/latency, I noticed that the redis-rdb-bgsave process is using up all the memory and swap space on the server. While looking through the redis-server.log, there are no errors shown but I see a lot of the below.
1301:C 23 Aug 12:42:00.554 * DB saved on disk
21301:C 23 Aug 12:42:02.498 * RDB: 934 MB of memory used by copy-on-write
1685:M 23 Aug 12:42:05.094 * Background saving terminated with success
1685:M 23 Aug 12:43:06.010 * 10000 changes in 60 seconds. Saving...
1685:M 23 Aug 12:43:08.422 * Background saving started by pid 21946
21946:C 23 Aug 12:57:03.793 * DB saved on disk
21946:C 23 Aug 12:57:05.678 * RDB: 852 MB of memory used by copy-on-write
1685:M 23 Aug 12:57:08.121 * Background saving terminated with success
1685:M 23 Aug 12:58:09.103 * 10000 changes in 60 seconds. Saving...
1685:M 23 Aug 12:58:11.566 * Background saving started by pid 22603
22603:C 23 Aug 13:11:58.770 * DB saved on disk
22603:C 23 Aug 13:12:00.661 * RDB: 1334 MB of memory used by copy-on-write
1685:M 23 Aug 13:12:03.196 * Background saving terminated with success
1685:M 23 Aug 13:13:04.042 * 10000 changes in 60 seconds. Saving...
1685:M 23 Aug 13:13:06.535 * Background saving started by pid 23458
23458:C 23 Aug 13:26:24.714 * DB saved on disk
23458:C 23 Aug 13:26:26.663 * RDB: 955 MB of memory used by copy-on-write
1685:M 23 Aug 13:26:29.260 * Background saving terminated with success
1685:M 23 Aug 13:27:30.052 * 10000 changes in 60 seconds. Saving...
1685:M 23 Aug 13:27:32.383 * Background saving started by pid 25127
25127:C 23 Aug 13:41:09.278 * DB saved on disk
25127:C 23 Aug 13:41:11.179 * RDB: 527 MB of memory used by copy-on-write
1685:M 23 Aug 13:41:13.714 * Background saving terminated with success
1685:M 23 Aug 13:42:14.008 * 10000 changes in 60 seconds. Saving...
1685:M 23 Aug 13:42:16.299 * Background saving started by pid 27140
I believe all this started because we had the server crash a while ago and Redis was not running for about a month. Now that I have started it Redis is trying to process all the backlog. See rdb files below.
/var/lib/redis# ls -lt -h
total 83G
-rw-rw---- 1 redis redis 36G Aug 23 14:08 temp-29298.rdb
-rw-rw---- 1 redis redis 41G Aug 23 13:55 dump.rdb
-rw-rw---- 1 redis redis 2.4G Aug 23 00:22 temp-1683.rdb
-rw-rw---- 1 redis redis 2.1G Aug 22 22:44 temp-8334.rdb
-rw-rw---- 1 redis redis 2.5G Jul 20 01:00 temp-8735.rdb
-rw-rw---- 1 redis redis 185M Apr 17 03:47 temp-30556.rdb
While looking through this Canvas group, there are a lot of different suggestions about flushing the Redis cached data and disabling "saving Redis to disk in redis.conf" by commenting out the save lines (i.e.: save 900 1). My questions about doing this: