Help Needed: Basic guide to use Redis to synchronise 2 instances of Blocky

[gspannu]

Tried searching on GitHub and elsewhere but could not find any guide for Redis and Blocky.

I am looking for a basic guide (what to write in the config.yml file) to have 2 instances of blocky synchronised

My Setup:

• Blocky Instance 1 running at Ip address: a.b.c.d
• Blocky Instance 2 running at IP address: w.x.y.z.
• Redis Server running at IP address: k.l.m.n

Could someone please guide me as to what the basic config.yml redis section needs to be to ensure that caching between the 2 instances is synced?

redis:
  address: k.l.m.n
  password: <my_password>
  database: 0
  required: false
  connectionAttempts: 10
  connectionCooldown: 3s

Will the redis section in both blocky instances be 100% identical? Or does each confg.yml for the Blocky instance have to be somehow specific?

Sorry... a newbie on the Redis stuff

[kwitsch]

If both blocky instances have identical Redis sections they sync their caches. 👍

[gspannu]

Thanks for the quick response

What all gets synced?

• Just the caches
• the blocking state (is this only done a global level or even at the individual group level, what happens if group names are different across the instances)
• pre-fetching cache

Also, is there any master/ slave in the sync relationship or do the caches sync across both instances?
How does the TTL of the cached entries work, when syncing - as each instance could have different min/max TTL specified in the config yaml?

How often/ frequency is the syncing done? If I clear (flush) the cache on 1 instance using the API, does it also clear the Redis entries (and thereby the cache on the 2nd instance as well)... or will the cash eon the 1st reload from the Redis/2nd instance instead?

Is there any writeup/ readme that explain the sync behaviour, expected operation and settings?

Sorry... lots of questions !

[kwitsch]

The caches are synced upon entry through a pub/sub channel and also stored with it's TTL.
If an instance is restarted the cache is retrieved from Redis.

Further the enabled state is also synced.

There is no master slave relationship or sync schedules since Redis takes care of it by itself.

[kwitsch]

Is there any writeup/ readme that explain the sync behaviour, expected operation and settings?

Nope, there's currently no detailed documentation regarding these details.

Expected behavior:

• all instances have the same cache entries
• the cache entries have identical TTLs
• all instances have identical blocking states

[gspannu]

I haven't fully understood all of the above yet (not your fault, its my own lack of understanding !) ... but many thanks for the details.

Q: If I clear (flush) the cache on the 1st instance using an API call, what will be the behaviour...

1. Option 1: It will clear the cache, sync with Redis, (and thereby also clear the cache on the 2nd instance as well)
2. Option 2: The cache on the 1st instance will be cleared, but rebuilt again from the Redis - thereby negating the flush cache operation !
3. Something else ?

[kwitsch]

3: the cache is flushed on instance one but isn't rebuilt. Instance 2 isn't affected.

[Kernald]

Related: how does DB (MySQL/Postgres) work for multiple instances? Is it safe to point all the instances to the same DB?