00:05:06 GMT <badboy_> jenia: why not both?
00:05:27 GMT <badboy_> jenia: Redis is a "datastructure server"
00:05:38 GMT <jenia> badboy_, buy it's in meory
00:05:54 GMT <badboy_> it can be used as a database, some datastructures it exposes can be used as a queue and it has pub/sub functionality as well
00:05:55 GMT <jenia> so when you shot down the system, the data is lost
00:06:01 GMT <badboy_> not necessarily
00:06:05 GMT <badboy_> Redis can persist to disk
00:06:15 GMT <jenia> badboy_, alright. cool thanks
00:06:36 GMT <jenia> but the stuff in the disk need to fit into RAM or be replicated somewhere I guess
00:07:10 GMT <badboy_> (even without persistence I'd call it a "database", there's nothing a database can't be ephemeral)
00:07:24 GMT <badboy_> yes, Redis always keeps all the data in ram
00:09:22 GMT <badboy_>  anto
00:09:23 GMT <badboy_> oops
07:43:53 GMT <Marbug> Hi, I got a 4 node sedis/sentinel setup, but when a node is down, it seems that around 1/4th of the time there are misses, so it seems that the node which went down is still visible by sentinel and sends requests to the node, but all replies are empty. Could it be a bug in PHP Predis installed through npm? Or it it a sentinel configuration as I found down-after-milliseconds being set to 60000, but this would mean that after 60s the
07:43:59 GMT <Marbug> problem shall be gone although it stays all the time. Despite of the fact that when the node tries to sync, an error occured after a minute or so, the node disapears as a slave and a while back it comes back. But the cach misses keeps on occuring.
07:44:58 GMT <Marbug> I also suspect that writes go to the master and reads from the slave, but since the sentinel probably says that the node which is running but not in a synced state is returned as a "good" node, which will then try to serve the requested key which will be empty as a result?
09:07:09 GMT <Marbug> nvm, it was a problem in one of the drivers, think I got it solved now :-)