Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster, 2015-05-30

| Channels | #gluster index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
01:15 CyrilPeponnet joined #gluster
01:15 CyrilPeponnet joined #gluster
01:15 CyrilPeponnet joined #gluster
01:16 CyrilPeponnet joined #gluster
01:30 tdasilva joined #gluster
01:45 nangthang joined #gluster
01:46 thangnn_ joined #gluster
01:55 plarsen joined #gluster
01:58 ira left #gluster
01:59 ira joined #gluster
02:11 plarsen joined #gluster
02:48 bharata_ joined #gluster
03:12 TheSeven joined #gluster
03:47 dusmant joined #gluster
03:52 kanagaraj joined #gluster
04:02 mcpierce joined #gluster
04:04 gem joined #gluster
04:12 kdhananjay joined #gluster
04:14 ira joined #gluster
04:22 spalai joined #gluster
04:36 prabu joined #gluster
04:47 hagarth joined #gluster
04:54 maveric_amitc_ joined #gluster
05:27 karnan joined #gluster
05:30 dusmant joined #gluster
05:48 glusterbot News from resolvedglusterbugs: [Bug 1177737] [AFR-V2] - Self-healing from mounts happens even when {metadata,entry,data}-self-heal is disabled. <https://bugzilla.redhat.co​m/show_bug.cgi?id=1177737>
05:50 hchiramm_ joined #gluster
05:58 TheCthulhu joined #gluster
06:03 TheCthulhu Does Gluster distribute read-loads? For example, for hosting a large volume of websites, I am using Gluster as the storage backend for the files, rather than a simple upgrade to SSD to reduce the load on an individual disk, plus give better HA. Would the file requests that are not caught in the nginx cache be distributed between the replicated volumes in a kind of round robin fashion or do it only use a particular one until it is unava
06:13 JoeJulian TheCthulhu: by default, it's first to respond which is almost always the first replica. You can configure that behavior. See "gluster volume set help"
06:13 JoeJulian cluster.read-hash-mode
06:17 TheCthulhu JoeJulian: Thanks. On the premise of first to respond, that would mean each file retrieval would cause load across all servers though correct, thus every replica will have to handle every request?
06:17 TheCthulhu So for a high volume of read requests, it will still bog down servers that aren't doing the read work if they didn't respond first
06:18 JoeJulian There's still going to be the health check on lookup either way.
06:18 JoeJulian So there has to be a query to every replica.
06:18 JoeJulian Unless you turn off self-heal checking.
06:19 JoeJulian Unless you're serving the same 1 file to 10000+ clients, you generally don't need very many replicas anyway.
06:20 JoeJulian If your serving a disparate set of files, the distribute translator will help you more.
06:21 JoeJulian https://joejulian.name/blog/glus​terfs-replication-dos-and-donts/
06:21 TheCthulhu Very true. My cluster is operating over LAN within 2ms of each differnet node in Gluster and the nginx webservers, and the load should be spread across pretty well with many files. Any commonly requested files will likely get caught by nginx too anyway, which has an SSD-based cache.
06:21 JoeJulian perfect
06:23 TheCthulhu Can it be set so a replica is made but not used by clients? Using it as a backup in an entirely different DC and country, and would be helpful to be able to gzip the whole archive without hitting performance on the replicas used for reading
06:23 JoeJulian That's georeplication
06:28 spalai joined #gluster
06:55 karnan joined #gluster
07:17 LebedevRI joined #gluster
07:23 verdurin joined #gluster
07:30 karnan joined #gluster
07:30 rjoseph joined #gluster
08:06 Leildin joined #gluster
08:47 Leildin joined #gluster
08:47 BoyDoy joined #gluster
08:53 ekuric joined #gluster
08:56 lexi2 joined #gluster
08:58 nangthang joined #gluster
09:15 ashiq joined #gluster
09:15 hgowtham joined #gluster
09:17 ProT-0-TypE joined #gluster
09:29 hgowtham joined #gluster
09:35 ashiq joined #gluster
09:44 Manikandan joined #gluster
10:49 tg2 joined #gluster
10:57 mcpierce joined #gluster
11:06 ramteid joined #gluster
11:23 Pupeno joined #gluster
11:49 gem joined #gluster
11:50 uebera|| joined #gluster
12:03 autoditac joined #gluster
12:29 kanagaraj joined #gluster
12:41 nsoffer joined #gluster
13:26 maveric_amitc_ joined #gluster
13:28 plarsen joined #gluster
13:37 mcpierce joined #gluster
13:39 elico joined #gluster
13:43 nsoffer joined #gluster
13:54 plarsen joined #gluster
14:08 elico joined #gluster
14:08 nsoffer joined #gluster
14:34 ProT-0-TypE joined #gluster
14:51 DV joined #gluster
14:55 nishanth joined #gluster
15:08 DV joined #gluster
15:31 DV joined #gluster
15:53 nangthang joined #gluster
16:16 ProT-0-TypE joined #gluster
16:19 jcastill1 joined #gluster
16:20 DV joined #gluster
16:25 jcastillo joined #gluster
16:31 DV joined #gluster
16:37 rjoseph joined #gluster
17:15 DV_ joined #gluster
17:32 soumya joined #gluster
17:44 wkf joined #gluster
18:40 jobewan joined #gluster
19:10 DV joined #gluster
19:27 jobewan joined #gluster
20:13 gem joined #gluster
20:43 gem joined #gluster
20:44 DJClean joined #gluster
20:45 Pupeno joined #gluster
20:45 Pupeno joined #gluster
20:51 Bardack joined #gluster
20:55 DJClean joined #gluster
20:57 rotbeard joined #gluster
21:03 barnim joined #gluster
21:17 DJClean joined #gluster
21:22 plarsen joined #gluster
21:30 Pupeno joined #gluster
21:54 Prilly joined #gluster
22:51 Pupeno joined #gluster
22:51 Pupeno joined #gluster
22:55 n-st joined #gluster
22:59 n-st joined #gluster
23:10 Pupeno joined #gluster
23:29 Prilly joined #gluster
23:48 Pupeno_ joined #gluster

| Channels | #gluster index | Today | | Search | Google Search | Plain-Text | summary