Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2017-06-20

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

All times shown according to UTC.

Time Nick Message
00:00 Alghost joined #gluster-dev
00:03 Alghost joined #gluster-dev
00:13 kkeithle joined #gluster-dev
00:13 xavih joined #gluster-dev
00:13 ndevos_ joined #gluster-dev
00:13 ndevos_ joined #gluster-dev
00:13 obnox_ joined #gluster-dev
00:13 misc_ joined #gluster-dev
00:13 nixpanic_ joined #gluster-dev
00:13 glustin joined #gluster-dev
00:13 nixpanic_ joined #gluster-dev
00:14 nigelb_ joined #gluster-dev
00:14 nigelb_ joined #gluster-dev
00:14 uebera|| joined #gluster-dev
00:14 uebera|| joined #gluster-dev
00:15 glusterbot joined #gluster-dev
00:15 JoeJulian joined #gluster-dev
00:16 kkeithle joined #gluster-dev
00:16 ndevos_ joined #gluster-dev
00:16 ndevos_ joined #gluster-dev
00:16 glustin joined #gluster-dev
00:16 Alghost joined #gluster-dev
00:20 purpleidea joined #gluster-dev
00:20 purpleidea joined #gluster-dev
00:23 Alghost joined #gluster-dev
00:24 decayofmind joined #gluster-dev
00:41 Alghost joined #gluster-dev
00:44 gyadav joined #gluster-dev
00:57 rastar_ joined #gluster-dev
01:04 shyam joined #gluster-dev
01:48 ilbot3 joined #gluster-dev
01:48 Topic for #gluster-dev is now Gluster Development Channel - https://www.gluster.org | For general chat go to #gluster | Patches - https://review.gluster.org/ | Channel Logs - https://botbot.me/freenode/gluster-dev/ & http://irclog.perlgeek.de/gluster-dev/
02:03 prasanth_ joined #gluster-dev
02:40 skoduri_ joined #gluster-dev
02:45 Alghost joined #gluster-dev
02:49 mchangir joined #gluster-dev
02:50 msvbhat joined #gluster-dev
03:07 msvbhat joined #gluster-dev
03:31 kotreshhr joined #gluster-dev
03:46 Alghost joined #gluster-dev
03:54 itisravi joined #gluster-dev
04:10 riyas joined #gluster-dev
04:10 ppai joined #gluster-dev
04:15 msvbhat joined #gluster-dev
04:19 susant joined #gluster-dev
04:20 Alghost joined #gluster-dev
04:25 Alghost_ joined #gluster-dev
04:29 atinm joined #gluster-dev
04:29 gyadav joined #gluster-dev
04:33 susant left #gluster-dev
04:34 jiffin joined #gluster-dev
04:47 sahina joined #gluster-dev
04:50 skumar joined #gluster-dev
04:51 pranithk1 joined #gluster-dev
05:03 prasanth joined #gluster-dev
05:07 aravindavk joined #gluster-dev
05:12 rastar_ joined #gluster-dev
05:12 msvbhat joined #gluster-dev
05:24 poornima joined #gluster-dev
05:24 kotreshhr left #gluster-dev
05:25 susant joined #gluster-dev
05:26 nbalacha joined #gluster-dev
05:32 nishanth joined #gluster-dev
05:33 ankitr joined #gluster-dev
05:33 rafi1 joined #gluster-dev
05:35 pkalever joined #gluster-dev
05:37 nishanth joined #gluster-dev
05:38 ankitr joined #gluster-dev
05:38 Shu6h3ndu joined #gluster-dev
05:40 karthik_us joined #gluster-dev
05:48 hgowtham joined #gluster-dev
05:49 apandey joined #gluster-dev
05:52 kdhananjay joined #gluster-dev
05:53 poornima joined #gluster-dev
05:55 Saravanakmr joined #gluster-dev
05:57 ashiq joined #gluster-dev
06:05 karthik_us joined #gluster-dev
06:25 nbalacha_ joined #gluster-dev
06:26 sona joined #gluster-dev
06:28 apandey joined #gluster-dev
06:36 ankitr joined #gluster-dev
06:53 skoduri_ joined #gluster-dev
06:56 Saravanakmr joined #gluster-dev
07:10 nbalacha_ joined #gluster-dev
07:10 susant left #gluster-dev
07:11 nbalacha joined #gluster-dev
07:30 msvbhat joined #gluster-dev
07:57 v12aml joined #gluster-dev
08:26 susant joined #gluster-dev
08:31 jiffin1 joined #gluster-dev
08:32 apandey_ joined #gluster-dev
08:38 csaba joined #gluster-dev
08:43 apandey__ joined #gluster-dev
08:45 ankitr joined #gluster-dev
08:45 susant joined #gluster-dev
08:46 atinm joined #gluster-dev
08:57 susant joined #gluster-dev
09:04 jiffin1 joined #gluster-dev
09:10 kdhananjay joined #gluster-dev
09:19 hgowtham joined #gluster-dev
09:20 poornima joined #gluster-dev
09:22 Saravanakmr joined #gluster-dev
09:23 rafi joined #gluster-dev
09:24 skoduri_ joined #gluster-dev
09:34 msvbhat joined #gluster-dev
09:41 hgowtham joined #gluster-dev
09:42 atinm joined #gluster-dev
09:47 Saravanakmr joined #gluster-dev
09:53 ppai joined #gluster-dev
09:54 major joined #gluster-dev
10:16 apandey_ joined #gluster-dev
10:19 apandey_ pkalever: we ra eon you BJ link
10:20 pkalever apandey_: do you want me to join ?
10:20 apandey_ pkalever
10:20 apandey_ pkalever: No
10:20 apandey_ pkalever: Sorry
10:20 pkalever apandey_: okay, you can use it
10:21 pkalever apandey_: do you mean to ping pranith ?
10:21 pkalever apandey_: I understand now :-)
10:26 rastar joined #gluster-dev
10:45 bfoster joined #gluster-dev
10:46 kdhananjay joined #gluster-dev
10:58 apandey joined #gluster-dev
11:02 mchangir joined #gluster-dev
11:03 Saravanakmr joined #gluster-dev
11:11 rraja joined #gluster-dev
11:16 Alghost joined #gluster-dev
11:16 skoduri_ joined #gluster-dev
11:18 ashiq joined #gluster-dev
11:22 rastar joined #gluster-dev
11:23 rraja_ joined #gluster-dev
11:33 pranithk1 xavih: hey, we are all waiting for your response on the geo-rep regression thread...
11:34 pranithk1 xavih: Aravinda suggested a simpler way to present the data..
11:38 pranithk1 aravindavk: xavih responded :-).
11:40 xavih pranithk1: yes. I think a simpler representation is easier to parse, but it's also less extensible when new needs arise...
11:40 xavih pranithk1: and then it's harder to find a way to make it backward compatible
11:41 pranithk1 xavih: okay. I don't have enough experience about these kinds of things, but to me, both ',' separated and () are similar for now. Both will need to be extended when we have to add more information.
11:42 pranithk1 xavih: did I miss something you know?
11:42 pranithk1 xavih: Please do add any detail, you think '()' will achieve that is not in ',' separated list where each AFR/EC subvols are separated by space
11:43 pranithk1 aravindavk: ^^
11:43 xavih pranithk1: if we simply do "U1 U2, U3 U4", what will happen when we need an additional layer because we create multiple DHT layers for example ?
11:43 pranithk1 aravindavk: we can just complete the conversation and post summary
11:44 aravindavk pranithk1: xavih Geo-rep is fine with any format if it can be distingushed as subvolumes
11:44 xavih pranithk1: with '()' it's easier to represent the hierarchy: "(U1, U2), (U3, U4)". Now if we have another layer: ((U1, U2), (U3, U4)), ((U5, U6), (U7, U8))
11:44 pranithk1 xavih: So are these '()' like in Lisp? If we have multiple dht layers it will be ...
11:44 pranithk1 xavih: you typed what I have in mind I think
11:44 xavih pranithk1: that is harder to represent only with spaces and ','
11:45 pranithk1 xavih: yeah got it!
11:45 xavih pranithk1: I think additional headers indicating the type of volume will be even better to make decisions and detect incompatible changes, but at least the '()' would be very useful. At least that's my opinion
11:46 pranithk1 xavih: looks good to me.
11:46 pranithk1 aravindavk: you are also fine with this approach?
11:47 pranithk1 aravindavk: does geo-rep work with tiering? In the sense are they tested?
11:47 xavih pranithk1: parsing support can be added in libglusterfs, so it's easier to use that information and make changes in the future
11:47 aravindavk pranithk1, xavih: In ((U1, U2), (U3, U4)), ((U5, U6), (U7, U8)) example, subvol seperator and record separator are same, parsing is difficult I guess
11:47 aravindavk pranithk1: yes, geo-rep works with Tiering
11:48 xavih aravindavk: there are '()' that delimits each subvolume
11:48 xavih aravindavk: inside each subvolume, ',' separates sub-sub-volumes
11:48 pranithk1 xavih: geo-rep is in python, so that logic will need to be added separately I think
11:48 xavih aravindavk: I can write the parsing functions if needed
11:49 xavih pranithk1: no problem. I can write it in python also
11:49 aravindavk xavih: I am fine working on parsing, no issues from Geo-rep side
11:49 pranithk1 xavih: :-), I meant we can't add it in libglusterfs.
11:50 xavih pranithk1: DHT will need it, and AFR and EC need to create it, so common functions in libglusterfs can be useful
11:50 pranithk1 skumar: karthik_us|afk: ^^ find susanth and tell him as well
11:50 pranithk1 xavih: ah! like that
11:50 xavih pranithk1, aravindavk: adding a prefix to each subvolume is not an option now ?
11:50 nbalacha joined #gluster-dev
11:51 mchangir joined #gluster-dev
11:51 pranithk1 xavih: I guess it is all because of parsing only.
11:51 xavih pranithk1, aravindavk: though node-uuid is probably not the best name for it...
11:52 xavih pranithk1: that's why I said I can write the parsing functions...
11:52 pranithk1 xavih: hmm... in that case shall we revert the old bahvior for node-uuid?
11:52 aravindavk xavih: prefix as in "DHT[2](AFR[2,0](NODE(U1), NODE(U2)), AFR[2,0](NODE(U1), NODE(U2)))" ?
11:52 xavih aravindavk: yes, that would tell a lot of info about the volume and make it possible to take more intelligent decissions
11:53 xavih aravindavk: and it would it allow to detect future incompatibilities and don't break backward compatibility
11:54 xavih aravindavk: for example if we expect AFR or EC, but XX is found, geo-rep can decide to not start any worker there. This way it won't cause harm on a subvolume of an unknown type
11:54 aravindavk xavih: backward compatibility is not an issue since this xattr is not exposed to user, but need to fix all components who are using this xattr
11:54 xavih aravindavk: any future addition that creates the XX type, will be safe
11:55 xavih aravindavk: it's not exposed to the user, but a single gluster cluster can have (during the upgrade procedure) multiple versions on each node
11:55 xavih aravindavk: so some node will understand a new feature while older ones won't
11:55 msvbhat joined #gluster-dev
11:55 aravindavk xavih: geo-rep already decides volume type by volinfo(subvol number, type of volume etc)
11:56 xavih aravindavk: ah, that's good
11:57 apandey susant: ^^
11:57 xavih aravindavk: ok, we can continue without the prefixes for now.
11:58 xavih aravindavk: hope that this approach is enough for future changes
11:59 aravindavk xavih: I think Geo-rep also can load libglusterfs as shared lib and use the parser function, I will check the possibility of that. If that is possible without any issues then we can have common parser in libglusterfs
11:59 skoduri_ joined #gluster-dev
12:00 xavih aravindavk: that would be great :)
12:05 pranithk1 xavih: aravindavk: So who is doing what now?
12:06 xavih pranithk1: I can do the parsing functions if we agree on the format (I would prefer the more verbose one... :P)
12:07 pranithk1 xavih: So you are writing both creation of the string in format and given a format get the gfids?
12:07 pranithk1 xavih: sorry uuids
12:08 xavih pranithk1: if you all are ok with that, yes
12:08 xavih pranithk1: I'm also thinking on doing some more advanced support like: return all unique UUID's from a node-uuid value
12:08 pranithk1 xavih: Can we get what is the first up gfid in an HA subvolume either EC/AFR one?
12:09 xavih pranithk1: of course
12:09 pranithk1 xavih: okay. How long will it take?
12:09 pranithk1 aravindavk: ^^
12:10 xavih pranithk1: that's a good question... I can work on it this afternoon and have something operational. To refine it and create more features maybe we'll need to wait till thursday afternoon...
12:11 pranithk1 xavih: We have deadline of Wednesday evening to get it merged upstream :-(
12:11 xavih pranithk1: :-/
12:12 xavih pranithk1: what if we go with the fast solution for now (i.e. using what aravindavk proposed) and implement the complete solution afterwards ?
12:12 pranithk1 xavih: looks good to me. aravindavk ^^
12:13 pranithk1 xavih: Since aravindavk said there is no compatibility problem we should be good
12:13 aravindavk pranithk1, xavih: good to me as well
12:13 pranithk1 xavih: okay so to summarize
12:13 xavih pranithk1: in this case, maybe we can create a new xattr, for example: 'layout'. And then we can progressively change users of node-uuid to layout
12:14 xavih pranithk1: maybe this way it will be easier to not break backward compatibility
12:14 pranithk1 xavih: You mean we will keep node-uuid to return just first up subvolume?
12:14 pranithk1 xavih: and create a separate xattr which returns multiple uuids as a separate one?
12:15 xavih pranithk1: no, no, this change is already done. I would modify it to add the ',' as a separator between subvolumes only
12:15 pranithk1 xavih: oh okay cool
12:15 xavih pranithk1: I think this is the faster way to fix the geo-rep problem
12:15 pranithk1 xavih: cool
12:15 xavih pranithk1: but layer we can add the 'DHT[x](...)' format as a new xattr
12:15 pranithk1 susant: start with ',' change in dht
12:15 pranithk1 xavih: but who will use it?
12:15 xavih s/layer/later/
12:16 xavih pranithk1: I think current users of node-uuid can use it to be more "intelligent" in their decisions
12:16 ashiq joined #gluster-dev
12:17 xavih pranithk1: for example, geo-rep could use this attribute to have all information and no need to check volinfo()
12:17 ashiq joined #gluster-dev
12:18 pranithk1 xavih: Alright. I think rebalance will have to enhance the current functionality to handle some edge conditions, may be that would be a good time for this extra stuff
12:18 xavih pranithk1: :)
12:18 pranithk1 xavih: so we have time for this one too.
12:19 pranithk1 xavih: I think Nithya is away today. May be we can have that conversation too and close on the final format. May be tomorrow
12:19 pranithk1 xavih: then we are good to go
12:19 xavih pranithk1: good
12:19 pranithk1 xavih: aravindavk: for now let us just change geo-rep and dht encode/parsing logic so that we come out of the problem
12:21 pranithk1 susant: could you ack please?
12:21 aravindavk pranithk1: sure, I will send georep patch to parse "U1 U2,U3 U4" format
12:21 pranithk1 aravindavk: thanks aravindavk! sorry for all the mess :-(
12:21 kdhananjay joined #gluster-dev
12:21 rastar joined #gluster-dev
12:21 pranithk1 aravindavk: this discussion should have happened before we sent the first patch. We should have sent the mail on gluster-devel
12:24 pranithk1 apandey: could you confirm this change from susant also?
12:24 pranithk1 apandey: s/from/to'
12:25 aravindavk pranithk1: I am logging out now, will send the georep patch tonight
12:25 pranithk1 aravindavk: cool
12:29 apandey pranithk1: ok
12:39 guest joined #gluster-dev
12:41 ndevos kkeithley: any comments on https://review.gluster.org/17339 ? if not, just +2 and merge it :)
12:42 kkeithley ndevos: done
12:42 ndevos kkeithley: oh, that was fast!
12:42 kkeithley <click><click>
12:45 kkeithley I'm taking your word that it's okay
12:46 kdhananjay joined #gluster-dev
12:47 skoduri_ joined #gluster-dev
12:49 susant pranithk1: +1
12:52 kdhananjay joined #gluster-dev
13:00 kotreshhr joined #gluster-dev
13:04 mchangir ndevos, kkeithley, so that was a click bait ? :)
13:05 kkeithley the best kind. ;-)
13:06 rraja_ joined #gluster-dev
13:06 mchangir kkeithley, boy, you should never be given to man a nuclear power plant :P
13:06 kkeithley Homer Simpson is my hero
13:06 mchangir :D
13:09 ndevos kkeithley: right, like that? http://stuffpoint.com/the-simpsons/image/46596-the-simpsons-the-simpsons-wallpaper2.jpg
13:11 kkeithley didn't know Homer could dance, but yeah
13:15 mchangir that reminds me: I still have to buy the entire collection of Tintin, Asterix and Obelix and The Simpsons
13:15 shyam joined #gluster-dev
13:16 atinm joined #gluster-dev
13:19 major joined #gluster-dev
13:33 msvbhat joined #gluster-dev
13:47 major joined #gluster-dev
13:59 Saravanakmr joined #gluster-dev
14:18 rafi joined #gluster-dev
14:35 ndevos aaargh, who thought that placing key shortcuts "close application" and "close tab" next to eachother was a good idea?
14:35 * ndevos hits CTRL+q but wanted CTRL+w :-(
14:49 aravindavk joined #gluster-dev
15:13 wushudoin joined #gluster-dev
15:18 pranithk1 joined #gluster-dev
15:19 kotreshhr left #gluster-dev
15:27 nbalacha joined #gluster-dev
15:30 atinm joined #gluster-dev
15:45 pranithk1 joined #gluster-dev
16:07 rastar joined #gluster-dev
16:14 msvbhat joined #gluster-dev
16:31 jiffin joined #gluster-dev
16:44 skumar joined #gluster-dev
16:47 msvbhat joined #gluster-dev
17:35 sona joined #gluster-dev
17:40 susant joined #gluster-dev
17:40 susant left #gluster-dev
17:55 sona joined #gluster-dev
18:26 ashiq joined #gluster-dev
19:00 vbellur joined #gluster-dev
19:09 rafi1 joined #gluster-dev
19:12 kotreshhr joined #gluster-dev
19:12 kotreshhr left #gluster-dev
19:19 rafi joined #gluster-dev
19:39 rafi joined #gluster-dev
20:43 shyam joined #gluster-dev
21:01 shyam joined #gluster-dev
21:52 shyam joined #gluster-dev
22:14 glustin_ joined #gluster-dev
22:14 glusterbot joined #gluster-dev
22:14 JoeJulian joined #gluster-dev
22:15 obnox joined #gluster-dev
22:15 nigelb joined #gluster-dev
22:15 xavih_ joined #gluster-dev
22:15 glustin_ joined #gluster-dev
22:16 [o__o] joined #gluster-dev
22:21 decayofmind joined #gluster-dev
22:36 shyam joined #gluster-dev
22:52 shyam joined #gluster-dev
22:53 shyam left #gluster-dev
23:04 Alghost joined #gluster-dev
23:51 Alghost joined #gluster-dev

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