Redis Shuts Down Every Other Week

alcoma

Hi,
We have had issues with Redis, intermittently for a while now - some errors solved, but this one not.
The main problem has been with Redis shutting down; I can easily restart it but I haven't been able to trace the cause and the redis logs haven't been very helpful. This is the error we see online when it happens:

_Fatal error: Uncaught RedisException: No route to host in /home/b3a245/public_html/pawtucket/ca/vendor/tedivm/stash/src/Stash/Driver/Redis.php: 124 Stack trace: #0 /home/b3a245/public_html/pawtucket/ca/vendor/tedivm/stash/src/Stash/Driver/Redis.php(124): Redis->connect(localhost', 6379,'1) #1 /home/b3a245/public_html/pawtucket/ca/vendor/tedivm/stash/src/Stash/Driver/AbstractDriver.php(39):Stash\Driver\Redis->setOptions(Array)#2/home/b3a24e5/public_html/pawtucket/ca/app/lib/Cache/ExternalCache.php(284):Stash\Driver\AbstractDriver-> construct(Array) #3 home/63a245/public_html/pawtucket/ca/app/lib/Cache/ExternalCache.php(228):ExternalCache::getRedisObjectO#4/home/b3a24e5/public_html/pawtucket/ca/app/lib/Cache/ExternalCache.php(57):ExternalCache::getCacheObject)#5/home/63a245/public_html/pawtucket/ca/app/lib/Cache/ExternalCache.php(171):
ExternalCache:init) #6 /home/b3a24e5/public_html/pawtucket/ca/app/lib/Cache/CompositeCache.php(86): ExternalCache::contains('setup_php_mtime', 'default) #7 /home/b3a24e5/public_html/pawtucket/ca/app/helpers/utilityHelpers.php(4001): CompositeCache::contains('setup_php_mtime") #8 /home/b3a245/public_html/pawtucket/ca/app/lib/Configuration.php(185):caSetupPhpHasChanged)#9/home/b3a24e5/public_html/pawtucket/ca/app/lib/Configuration.php(109):Configuration-> construct('/home/b3a24e5/p...', true, false, true) #10
home/63a245/public_html/pawtucket/ca/app/lib/Configuration.php(162): Configuration:load( home/b3a245/p...', false, false, true) #11 /home/63a245/public_html/pawtucket/ca/app/lib/Configuration.php(109): Configuration->construct(/home/b3a24e5/p...', true, false, false) #12
home/b3a245/public_html/pawtucket/ca/app/lib/Configuration.php(166): Configuration::load(/home/b3a24e5/p..) #13 /home/b3a24e5/public_html/pawtucket/ca/app/lib/Configuration.php(109): Configuration->construct(home/b3a24e5/p...', true, false, false) #14 home/63a245/public_html/pawtucket/ca/app/helpers/utilityHelpers.php(63):Configuration::load(/home/b3a245/p...)#15/home/b3a24e5/public_html/pawtucket/ca/app/lib/Media/MedialnfoCoder.php(39):require_once('home/b3a24e5/p...#16/home/b3a24e5/public_html/pawtucket/ca/app/lib/Db/DbResult.php(33):
include_once('home/b3a24e5/p...') #17 /home/b3a24e5/public_html/pawtucket/ca/app/lib/Db/DbStatement.php(39):require_once('/home/b3a24e5/p...)#18/home/b3a24e5/public_html/pawtucket/ca/app/lib/Db.php(38):require_once('/home/b3a24e5/p...)#19 home/b3a245/public_html/pawtucket/ca/app/lib/Cache/PersistentCache.php(33):require_once('/home/b3a245/p...#20/home/b3a24e5/public_html/pawtucket/ca/app/helpers/preload.php(82):require_once('home/b3a24e5/p...)#21/home/b3a24e5/public_html/pawtucket/ca/app/helpers/post-setup.php(372):
require_once('/home/b3a245/p...') #22 /home/b3a24e5/public_html/pawtucket/ca/setup.php(250): require('/home/b3a24e5/p...) #23 /home/b3a24e5/public.
html/pawtucket/ca/index.php(37): require('/home/b3a24e5/p...') #24 {main} thrown in
/home/b3a245/public_html/pawtucket/ca/vendor/tedivm/stash/src/Stash/Driver/Redis.phpon line 124

Everything works fine when restarting redis but we'd like to understand what is causing it to stop in the first place and our web host support can't tell us much as these errors ref C/A. Thanks.

seth

It may be crashing. I'd check the redis.conf maxmemory and maxmemory-policy settings. maxmemory should be set to a value that fits within available memory. maxmemory-policy should be set to something that evicts cache entries when the maximum memory limit is reached.

alcoma

Thanks. It was increased from 8 to 12GB so let's see if that helps.

monica

alcoma how much RAM does your server have? 12gb for Redis on a shared server seems a bit high. When this happens (redis crashing), check the syslogs for evidence of an OOM killer trigger. Your system might have run out of RAM and if redis is the service using the most RAM it is likely picked to be killed to protect the system.

alcoma

Thanks Monica. We have a private vps with 20gb of ram. So yes, it seems we are abundant enough with memory.
When looking at the current processes, I do see redis using an unusual amount of the cpu and memory:

redis 0(priority) 84.20 (cpu%) 27.94 (memory%) redis-rdb-bgsave 127.0.0.1:6379

seth

I'd the maxmemory lower and make sure you have an eviction policy set.

alcoma

Right. I did check the redis memory in the cli and noticed that the eviction policy was noeviction so I changed it to allkeys-lru.
I'm going to set the maxmemory back to 8gb; I figure that is a fair level given our dataset is approx. 2gb not including media ?

seth

8gb is still quite high. 2gb would be a high, but not insane. The eviction policy is likely most significant though. See how it goes now that it's changed.

alcoma

One more question:
What has been your experience with settings for RDB snapshots ?
And preference of redis over memcached ? Considering switching.