15:07:26 GMT <nnog> Anyone here run into an issue with trying to keep sentinel networks separate and sometimes replicate between them without the sentinels learning about temporary slaves?
15:08:46 GMT <nnog> My problem is I sometimes want to do SLAVEOF to replicate (effectively copy the data) from one server to another server, but I have sentinels for discovery and uptime that also want to enforce slave config
15:09:17 GMT <nnog> I want to somehow prevent them from enforcing this slave config for when I want to temporarily slave and unslave these servers
15:09:48 GMT <nnog> any ideas? Ideally a way to do it without modifying and rebuilding sentinel.c
15:10:59 GMT <nnog> I was thinking a hacky way could be to rename the SLAVEOF and/or the CONFIG REWRITE commands in the servers so the sentinels are unable to "correct" the replication configs
15:17:27 GMT <minus> i know the problem but do not have a solution
21:07:24 GMT <endpoint_david> hi folks, does GETSET/SET block other concurrent accesses for the same key while setting?
21:07:54 GMT <endpoint_david> obviously the choice would seem to be return value old key or block; in my case old data is preferred
21:08:42 GMT <endpoint_david> since the set is atomic, I would expect that it just blocks, but would like some confirmation
21:11:59 GMT <minus> every redis command blocks; redis is single threaded
21:16:01 GMT <endpoint_david> ok, good to know. :-)  thanks for the info