15:13:32 GMT <rburton> re
15:13:48 GMT <rburton> Is it allowed to put an expire on a single element within RPOPLPUSH
15:15:05 GMT <rburton> Answer is no - okay now how to resolve this
18:53:01 GMT <wCPO> I currently have around ~ 200 Linux boxes which "ping" my backend roughly every 5 minute, I want to store the last 100 pings and expire pings after 14 days, looking through the datatypes, list seems like the ideal solution but then I would need a list per box, is that a bad idea?
18:56:40 GMT <minus> using a list per box and trimming it to 100? sounds like a plan
18:57:18 GMT <minus> 200 lists with each 100 entries is nothing, if you're asking that
18:58:40 GMT <hkdsun> That shouldn't even be a blip :) I've seen thousands of push/pops per second on tens of lists
18:59:31 GMT <wCPO> minus, and then just some sort of "cronjob" which remove old "pings", I can only use the EXPIRE functionally for the whole list as I understand.
19:00:45 GMT <minus> wCPO: correct
19:00:59 GMT <minus> you could also just have the pinging clients do that
19:02:13 GMT <wCPO> I'm mostly asking because I got a bit "scared" by this: http://www.mikeperham.com/2015/09/24/storing-data-with-redis/ , but prefixing every list with ex: box- so the list name would be box-<mac address>, should work I think.
19:02:28 GMT <minus> https://redis.io/commands/ltrim there's even an example about that there
19:03:51 GMT <wCPO> minus, the box could vanish without warning, so I still need something to cleanup "vanished" boxes, so some sort of cronjob seems easiest.
19:04:41 GMT <minus> you could set an expire on the whole list for that
19:06:09 GMT <wCPO> minus, but looping through the list every 5 minute when the client "ping" seems a bit overkill, running a cronjob once six 6 seems more optimal performance wise.
19:06:25 GMT <hkdsun> Could he use a sorted set with timestamps as scores? and then trim based on the age (timestamp) of members (pings)?
19:06:48 GMT <minus> regarding that article: that's about storing different kinds of data mainly, e.g. sessions and persistent data. if you don't store multiple types in your instance it doesn't affect you anyway
19:07:22 GMT <wCPO> hkdsun, I looked a bit on that, but I don't understand how I could get the last 100 pings for X box?
19:07:32 GMT <wCPO> minus, I see now :)
19:07:51 GMT <minus> wCPO: LTRIM would just trim the list to 100 entries, nothing with age. for age-based trimming hkdsun's suggestions works best
19:09:25 GMT <wCPO> The current solution is based on PHP and MySQL and it works, but there currently isn't anything cleaning up the table. So this is mostly rewriting for fun (and I have a few other "parts" where I also want to use redis), I don't really need the performance, but hopeful I will at some point.
19:11:17 GMT <wCPO> minus, but how can I get ex the last 100 pings from X box, with hkdsun approch? I don't really get that part..
19:11:29 GMT <minus> trimming a sorted set to 100 elements would work with ZREMRANGEBYRANK and trimming it to 14 days by ZREMRANGEBYSCORE
19:11:40 GMT <hkdsun> ZRANGE box-X-pings -100 -1 should give you the 100 most recent pings if your scores are timestamps
19:12:00 GMT <hkdsun> double-check my command though :)
19:12:30 GMT <wCPO> hkdsun, oh yeh, a sorted set per box. I had one big sorted set in my head :)
19:12:32 GMT <hkdsun> and minus' suggestion for trimming
19:15:09 GMT <wCPO> A list with ltrim and expire on "ping" seems like the most simple solution. I could end up with a list with 99 > days old ping, but I don't care too much about that.
19:15:20 GMT <wCPO> 99 > 14 old pings*
19:17:17 GMT <wCPO> Thanks for the inputs both of you :)
19:17:58 GMT <hkdsun> good luck!
20:31:57 GMT <wrkrcoop> hello. im trying to connect to my redis cluster (elastic cache service) on aws but im getting this error: Error: Redis connection to d2-.usw1.cache.amazonaws.com:6379 failed - connect ETIMEDOUT 1xx.xx.xx.xxx:6379
20:32:08 GMT <wrkrcoop> i figure ppl in here have used redis with AWS, anyone know what this could mean?