Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2015-07-29

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

All times shown according to UTC.

Time Nick Message
00:01 shyam joined #gluster-dev
01:47 ilbot3 joined #gluster-dev
01:47 Topic for #gluster-dev is now Gluster Development Channel - http://gluster.org | For general chat go to #gluster | Patches - http://review.gluster.org/ | Channel Logs - https://botbot.me/freenode/gluster-dev/ & http://irclog.perlgeek.de/gluster-dev/
02:07 shyam joined #gluster-dev
02:09 gem joined #gluster-dev
03:08 Gaurav__ joined #gluster-dev
03:27 badone joined #gluster-dev
03:33 nishanth joined #gluster-dev
03:34 kdhananjay joined #gluster-dev
03:38 kanagaraj joined #gluster-dev
03:42 atinm joined #gluster-dev
03:49 vmallika joined #gluster-dev
03:52 nbalacha joined #gluster-dev
03:52 shubhendu joined #gluster-dev
04:03 badone joined #gluster-dev
04:09 ppai joined #gluster-dev
04:09 hagarth joined #gluster-dev
04:17 kotreshhr joined #gluster-dev
04:20 nbalacha joined #gluster-dev
04:20 gem joined #gluster-dev
04:27 byreddy_ joined #gluster-dev
04:33 ndarshan joined #gluster-dev
04:33 overclk joined #gluster-dev
04:38 hchiramm joined #gluster-dev
04:43 pppp joined #gluster-dev
04:43 hagarth joined #gluster-dev
04:50 apandey joined #gluster-dev
04:52 rafi joined #gluster-dev
04:54 sakshi joined #gluster-dev
04:57 vikumar joined #gluster-dev
05:04 krishnan_p joined #gluster-dev
05:16 ashiq joined #gluster-dev
05:17 Manikandan joined #gluster-dev
05:17 hgowtham joined #gluster-dev
05:28 vmallika joined #gluster-dev
05:31 deepakcs joined #gluster-dev
05:34 hchiramm joined #gluster-dev
05:40 nbalacha joined #gluster-dev
05:40 Gaurav__ joined #gluster-dev
05:40 anekkunt joined #gluster-dev
05:42 Bhaskarakiran joined #gluster-dev
05:45 atalur joined #gluster-dev
05:45 rafi joined #gluster-dev
05:48 shubhendu joined #gluster-dev
05:49 ndarshan joined #gluster-dev
05:59 kdhananjay joined #gluster-dev
06:01 kshlm joined #gluster-dev
06:04 aravindavk joined #gluster-dev
06:10 Saravana_ joined #gluster-dev
06:14 kshlm hagarth, you around?
06:14 kshlm How do you post the release-announcement to the blog, facebook, g+ etc.?
06:18 gem joined #gluster-dev
06:20 anekkunt joined #gluster-dev
06:28 Manikandan joined #gluster-dev
06:33 raghu joined #gluster-dev
06:36 anoopcs Is there anybody facing issues with review.gluster.org ? I got the following error on git pull
06:36 anoopcs fatal: internal server error
06:37 anoopcs fatal: protocol error: bad pack header
06:37 nbalacha I see the same error
06:39 rafi anoopcs: me also :(
06:40 anoopcs kshlm, If possible, can you please check for any gerrit issues with review.gluster.org?
06:42 kshlm We need to run git gc on the server.
06:42 kshlm Are either hchirram or hagarth around?
06:42 kdhananjay anoopcs: Me too!
06:46 rafi problem is with http protocol, you can still use git or ssh
06:48 anoopcs kshlm, hchirram is on PTO.
06:51 skoduri joined #gluster-dev
06:52 anoopcs rafi, git is working for me but not ssh.
06:53 rafi anoopcs: ok,
06:54 jiffin joined #gluster-dev
06:54 anoopcs rafi, Most of us will be using ssh keys is what I guess.
06:57 ndarshan joined #gluster-dev
07:01 kshlm anoopcs, rafi, kdhananjay, nbalacha, can you guys try now.
07:01 anoopcs kshlm, sure.
07:01 kdhananjay atinm: rastar_afk hagarth: Could one of you please merge http://review.gluster.org/#/c/11776/ ?
07:02 kdhananjay atinm: rastar_afk hagarth I seems to be failing on every patch in 3.7.
07:02 kdhananjay s/I/It/
07:02 anoopcs kshlm, It's working fine now.
07:02 anoopcs at least for me
07:03 kdhananjay Working now, kshlm. kshlm++
07:03 glusterbot kdhananjay: kshlm's karma is now 21
07:04 anoopcs kshlm++
07:04 atinm kdhananjay, kshlm is merging it I believe as I see a +2 from him
07:04 glusterbot anoopcs: kshlm's karma is now 22
07:05 shubhendu joined #gluster-dev
07:05 kdhananjay atinm: OK.
07:05 rafi kshlm++
07:05 glusterbot rafi: kshlm's karma is now 23
07:16 byreddy_ joined #gluster-dev
07:31 sakshi joined #gluster-dev
07:35 kshlm kdhananjay, Can you check if the change is merged? I pressed submit, but gerrit conked out after that.
07:36 kdhananjay kshlm: Looking ...
07:36 kdhananjay kshlm: Yep. Merged!
07:54 sakshi joined #gluster-dev
08:23 byreddy_ joined #gluster-dev
08:30 pranithk joined #gluster-dev
08:47 sakshi joined #gluster-dev
08:58 ndevos hagarth, atinm, rafi, overclk, kshlm, *: who's hosting todays community meeting?
08:58 atinm ndevos, I am not as I will be leaving early
08:59 kdhananjay joined #gluster-dev
08:59 ndevos atinm: okay, lets see if there is someone else who can do it?
08:59 ndevos kdhananjay: interested in hosting todays community meeting?
08:59 rafi ndevos: I don't have enough experience to host community meeting ;)
09:00 kdhananjay ndevos: same as above ;)
09:00 ndevos rafi: uh, you have more than enough, the bug triage meetings are going well!
09:00 ndevos kdhananjay: you and rafi can do it together :)
09:01 kdhananjay rafi: Is that okay? :)
09:01 kshlm ndevos, I can do it.
09:02 kshlm But if kdhananjay or rafi volunteer, I'll be happy to let them do it.
09:03 rafi kdhananjay: ndevos : Ok , we will do it for next community meeting
09:03 rafi kdhananjay: what do you think ?
09:03 ndevos kshlm: yeah, I'd like to see many different people hosting the meeting
09:03 kdhananjay rafi: Agree with you.
09:03 kshlm rafi, the next meeting is today.
09:03 ndevos rafi, kdhananjay: okay, I'm in your office next week, so we can even sit together if that helps
09:03 kshlm Or did you mean next week's meeting?
09:04 kdhananjay ndevos: Sounds good. :)
09:04 rafi kshlm: yes, we think, the need some more coverage to it ;)
09:04 rafi kshlm: we will closely watch todays meeting :)
09:05 anekkunt atinm, have a look into this patch http://review.gluster.org/#/c/10850/
09:05 ndevos kshlm: so, if you would host it today, that would be good :)
09:05 rafi ndevos: kdhananjay : great
09:07 ndevos rafi, kdhananjay: thanks, and put it in your agenda :)
09:09 rafi ndevos: sure, :)
09:14 kdhananjay ndevos: Sure. :)
09:32 anekkunt joined #gluster-dev
09:35 kdhananjay1 joined #gluster-dev
09:57 pranithk all: Who is tracking latest spurious failures? I would like to if no one else is...
09:57 pranithk ndevos: atinm: rafi: ^^
10:00 ashishpandey joined #gluster-dev
10:01 wushudoin joined #gluster-dev
10:01 ndarshan joined #gluster-dev
10:01 KennethDejonghe joined #gluster-dev
10:03 mikedep333 joined #gluster-dev
10:03 akay1 joined #gluster-dev
10:04 ndevos pranithk: I'm not sure, did you update https://public.pad.fsfe.org/​p/gluster-spurious-failures or was that someone else?
10:05 Manikandan_ joined #gluster-dev
10:05 pranithk joined #gluster-dev
10:06 ndevos [REPEAT] pranithk: I'm not sure, did you update https://public.pad.fsfe.org/​p/gluster-spurious-failures or was that someone else?
10:06 ndevos kshlm: if you are hosting the meeting in ~2 hours, could you send a reminder to the list if you have not done so yet?
10:07 overclk joined #gluster-dev
10:08 suliba joined #gluster-dev
10:08 rafi pranithk: I'm also interested , ;)
10:11 pranithk ndevos: rafi: I just sent a mail on gluster-devel. RaSTar and I will be following up
10:14 ndevos pranithk: ok, sounds good to me
10:21 byreddy_ joined #gluster-dev
10:32 anekkunt joined #gluster-dev
10:40 ira_ joined #gluster-dev
10:56 hgowtham joined #gluster-dev
11:00 samikshan joined #gluster-dev
11:01 tdasilva joined #gluster-dev
11:16 firemanxbr joined #gluster-dev
11:27 shubhendu joined #gluster-dev
11:28 shyam joined #gluster-dev
11:31 kshlm Weekly community meeting starts in 30 minutes on #gluster-meeting. Agenda: https://public.pad.fsfe.org/​p/gluster-community-meetings
11:53 overclk joined #gluster-dev
12:01 kshlm Weekly community meeting starting now in #gluster-meeting
12:01 ppai joined #gluster-dev
12:20 kanagaraj joined #gluster-dev
12:21 kaushal_ joined #gluster-dev
12:28 jrm16020 joined #gluster-dev
12:31 spalai joined #gluster-dev
12:34 spalai joined #gluster-dev
12:39 spalai left #gluster-dev
12:47 hchiramm joined #gluster-dev
12:53 ppai joined #gluster-dev
12:57 kotreshhr left #gluster-dev
13:06 pranithk ndevos: dude, I wanted to talk to you about gf_ref_t and integrating mem-accounting kind of infra for it...
13:07 ndevos pranithk: when, now?!
13:07 pranithk ndevos: depends, are you up for it?
13:08 pranithk ndevos: I had a very bad time for the past two times because of ref leaks. Hated the experience :-(
13:08 ndevos pranithk: uh, barely... I've not seen the mem-accounting lately, so I would need to refresh that a little
13:09 pranithk ndevos: its simple. You tell how many types of allocations will be done.
13:09 pranithk ndevos: then you keep using the type of allocation at the time of allocation
13:09 pranithk ndevos: I am wondering how easy it would be to incorporate that in gf_ref_t
13:10 ndevos pranithk: what are we counting?
13:11 ndevos pranithk: the number of types is static per xlator, an allocation is done once, and free'd once... where's the refcount?
13:11 pranithk ndevos: every inode_ref/inode_unref will give a tag which will tell who is reffing/unreffing
13:11 pranithk ndevos: got it?
13:12 ndevos pranithk: uh, nope, where does inode_ref/inode_unref come in?
13:13 pranithk ndevos: we are going to change inode to use gf_ref_t :-D
13:13 ndevos pranithk: oh, sure, thats a no-brainer
13:13 pranithk ndevos: we will need to make some changes. But that is the plan...
13:14 pranithk ndevos: here is what I have observed. When there are ref-leaks. First thing to find is which component is causing it
13:14 pranithk ndevos: next thing would be to see which code path calls ref/unref on the object
13:15 ndevos pranithk: yep, I'm still following
13:15 pranithk ndevos: here is what I am thinking...
13:16 pranithk ndevos: lets say we have dictionary in inode_t which can keep track of refs/unrefs count.
13:16 ndevos pranithk: ok
13:16 pranithk ndevos: we will give REF/UNREF macros which take the current function which call ref/unref
13:17 ndevos pranithk: sure
13:17 pranithk ndevos: in the inode_t's dictionary we will put this information. i.e. if the function exists increment it, if it doesn't then decrement it
13:17 pranithk ndevos: chi chi
13:17 pranithk ndevos: wrong wrong
13:18 ndevos pranithk: you lost me again, we give inode_t a dict with functions in it?
13:18 pranithk ndevos:  in the inode_t's dictionary we will put this information. i.e. in REF we increment the count against the function-name and in unref we decrement the count against function name
13:18 ndevos pranithk: ah, I now think I get it!
13:19 pranithk ndevos: no, inode_t has a dict. In which we keep the information saying on this function we did these many increments. and on these functions we did so many decrements. We can even keep separate dicts for ref/unref but I feel it is too much book keeping...
13:19 ndevos pranithk: you want to count the functions that do the gf_ref/gf_unref, why not put that in gf_ref_t for a debug build?
13:20 pranithk ndevos: You got the idea right? finding the ref leak is so easy. You will just look at the functions and counters and find who the culprit is
13:20 pranithk ndevos: it would be better to put it in gf_ref_t so that all structures which use it can benefit automagically :-D
13:21 ndevos pranithk: yeah, I get your idea, but checking in a dict is quite some overhead, isnt it?
13:21 pranithk ndevos: That is the data-structure that is readily available which does the job for us....
13:21 ndevos rafi, jiffin: a weekend trip to Kerala, is that doable?
13:21 ndevos ah, wrong window!
13:22 ndevos pranithk: readily available, in inode_t or gf_ref_t?
13:23 ndevos pranithk: if we add this functionality, we should probably make it usable for all gf_ref_t users
13:24 pranithk ndevos: I meant dict_t is readily available :-)
13:25 pranithk ndevos: wait! in statedump, we can print these data structures. At least inode_t and fd_t can be printed.
13:26 ndevos pranithk: you mean inode_t already has a dict? we could pass an optional dict to GF_REF_INIT() or whatever the function is called
13:27 ndevos pranithk: that means the actual work would be done in refcount.c, other structures can use it, and can opt-in
13:27 pranithk ndevos: we will add a dict to gf_ref_t
13:27 pranithk ndevos: I think we are thinking the same thing :-)
13:28 ndevos pranithk: okay, sounds good then!
13:28 pranithk ndevos: so design is frozen?
13:29 pranithk ndevos: brb
13:30 ndevos pranithk: yes, if the design is: 1. add a dict to gf_ref_t, 2. GF_REF() adds calling function to the new dict and +1 it, 3. GF_UNREF() -1 the calling function in the dict, 4. only enabled in debug builds
13:30 shyam joined #gluster-dev
13:34 pranithk ndevos: 5. statedumps can use it
13:35 ndevos pranithk: okay, yes, add a function refcount.c that helps with the statedump bit
13:35 ndevos APPROVED
13:35 pranithk ndevos: cool!
13:36 pranithk ndevos: Let me see if any of our interns is interested in doing this. Or may be prasanna
13:36 ndevos pranithk: sure, I wasnt assuming you wanted to leave the coding to me ;-)
13:37 pranithk ndevos: well we can do it. But it would be nice if the new guys get their hands dirty. They will also feel they achieved something.
13:38 ndevos pranithk: yeah, find someone, discuss the idea, have it sent to the -devel list and start the coding after that
13:39 pranithk ndevos: yes sir
13:39 ndevos pranithk: great plan, thanks for bringing it up and have me think about it :D
13:40 * ndevos desperately needs a coffee now
13:41 overclk joined #gluster-dev
13:46 * pranithk leaves for home.
13:46 pranithk ndevos: cya dude
13:46 pranithk left #gluster-dev
14:08 nishanth joined #gluster-dev
14:13 pousley joined #gluster-dev
14:33 dlambrig1 joined #gluster-dev
14:48 nbalacha joined #gluster-dev
15:01 hagarth joined #gluster-dev
15:02 overclk joined #gluster-dev
15:08 vimal joined #gluster-dev
15:13 kshlm joined #gluster-dev
15:19 skoduri joined #gluster-dev
15:22 _Bryan_ joined #gluster-dev
15:56 ashiq joined #gluster-dev
15:58 ashiq joined #gluster-dev
16:08 sankarshan_ joined #gluster-dev
16:14 Gaurav__ joined #gluster-dev
17:06 overclk joined #gluster-dev
17:24 jiffin joined #gluster-dev
17:27 ashiq joined #gluster-dev
17:29 cholcombe joined #gluster-dev
17:29 cholcombe anyone have a sec to talk about gluster rpc?
17:29 cholcombe ndevos maybe?
17:30 cholcombe kkeithley?
17:30 kkeithley what about it?
17:30 kkeithley what about rpc?
17:30 cholcombe so i'm trying to understand how the cli rpc mechanism works
17:30 cholcombe i'm using the gluster pool list command just as a test case because it's small
17:31 cholcombe it seems that it xdr serializes 2 fragments.  the request and then the gf_cli_req second
17:31 cholcombe is that correct?
17:31 cholcombe the first fragment is sun rpc specific.  it has the prog, vers, xid, etc
17:32 cholcombe it then says the next part is procedure specific.  I don't know if I'm supposed to send a call header with that or just send the fragment itself
17:32 kkeithley It's been a long time since I looked at the nuts and bolts of sunrpc, including ours. That seems basically correct
17:32 kkeithley dunno the answer to the second question
17:32 cholcombe i think i'm close but on the glusterd side it's saying my msg payload is incorrect.
17:33 cholcombe it's weird because it gets all the way to here: __glusterd_handle_cli_list_friends
17:33 cholcombe and then bails on xdr_to_generic
17:33 cholcombe so my rpcsvc_request_t must be bogus but i'm not sure how to fix it :-/
17:34 kkeithley you're just trying to add a new command?
17:34 cholcombe yup
17:34 cholcombe so i'm trying to work out the details of how a simple command works
17:35 cholcombe do i send a dict to the server to fill ?
17:35 cholcombe i guess that's what i'm wondering
17:35 cholcombe the dict just seems to be a struct with a length and a string for the cli_req
17:37 jobewan joined #gluster-dev
17:38 hagarth cholcombe: have you added your cli xdr representation in cli1-xdr.x?
17:38 kkeithley the dict should be an array of name-value strings. Maybe an array of one, but it should still be an array
17:38 cholcombe ok cool
17:39 cholcombe hagarth: not yet.
17:40 hagarth cholcombe: you would just need to add that to .x file. once done xdrgen would take care of autogeneration of .c & .h files.
17:40 cholcombe ok i seem to be missing the xdrgen command.  I have rpcgen.  is that the same thing?
17:41 cholcombe i think ubuntu defaulted to installing libtirpc
17:41 hagarth cholcombe: yes, they are the same
17:41 cholcombe ok cool
17:42 hagarth all xdr.[ch] files are generated by rpcgen
17:42 cholcombe yeah i wish i had known that a week ago ;)  That would've saved me a ton of time
17:43 cholcombe i kept tracing through the code going what?  where does this come from and ctags couldn't find it
17:43 hagarth cholcombe: I sometimes view xdr as a blackbox, seldom try to make sense of the auto generated code ;)
17:43 cholcombe lol
17:43 cholcombe i know what you mean.  it's pretty hard to sort out what is going on
17:44 jiffin joined #gluster-dev
17:44 cholcombe hagarth: my ultimate goal is to create an API that uses rpc calls to give full api control over gluster
17:45 cholcombe since xdr is language agnostic it should be possible
17:47 hagarth cholcombe: right, that should be very much possible
17:47 cholcombe it's something i've wanted for years but since it hasn't shown up yet I figured I might give it a go
17:48 hagarth cholcombe: have you seen the APIs being defined for heketi/glusterd-2.0?
17:48 cholcombe i haven't
17:48 cholcombe are they just wrapping the cli like all the others?
17:49 cholcombe hagarth: oh yes I seem to remember this.  I think it was just a cli wrapper
17:49 hagarth cholcombe: no, the intent is to provide completely ResTful interfaces
17:49 hagarth cholcombe: https://github.com/heketi/heketi/wiki/API
17:49 cholcombe damn.. so you're already working on this haha
17:49 cholcombe i'm always a day late to these things
17:50 hagarth cholcombe: heketi intends providing higher level interfaces that would be useful for storage as a service scenarios
17:50 cholcombe interesting.  it's written in go
17:50 cholcombe hagarth: that's exactly why I need it.  I'm building storage as a service code
17:51 hagarth cholcombe: awesome, maybe we could collaborate on heketi/glusterd-2.0
17:51 cholcombe i'd be happy to
17:51 cholcombe i don't know Go but i'd learn it to contribute
17:52 cholcombe hagarth: it looks similar enough to rust that I could prob pick it up quickly
17:52 hagarth great, there was a related thread on gluster-devel about review of heketi APIs. Feel free to drop a note if you see the node for more APIs.
17:52 hagarth cholcombe: yes, Go is quite easy to pick up
17:52 cholcombe so eventually the CLi wrapper will go away in this and it'll use the native rpc's?
17:52 cholcombe because i saw there is xdr support in Go
17:54 hagarth cholcombe: right, glusterd would be a restful server with Heketi. The cli eventually would talk ResT I think.
17:54 cholcombe interesting
17:54 cholcombe sounds good to me.  I'll see if I can sign myself to build some of the pieces it needs
17:55 hagarth kshlm might have more details on how this will evolve
17:55 cholcombe ok
17:55 hagarth cholcombe: great, that would be fantastic. storage as a service readiness is the prime goal for Gluster.next releases (read 3.8 and 4.0).
17:55 cholcombe yeah i think it's sorely needed
17:58 hagarth indeed. In my conversations with various users, I have seen different mechanisms being used to provide this service with Gluster. It makes more sense for gluster to provide building blocks so that deploying a solution will be far more easier.
17:58 cholcombe hagarth: yeah i agree
18:00 cholcombe hagarth: so the plan is to land this work in september?
18:02 hagarth cholcombe: around then, before the october release of openstack IIUC
18:02 dlambrig1 left #gluster-dev
18:03 cholcombe ok great.  that corresponds well with my cycle that follows openstack
18:13 hagarth cholcombe: it might be better to check the exact plan on gluster-devel. lpabon et al wanted to have the basic APIs needed for Manila going before the next OpenStack release.
18:14 cholcombe hagarth: already I'll send them an email asking how I can help
18:15 hagarth cholcombe: cool, ok.
18:15 cholcombe thanks for the help
18:20 shyam joined #gluster-dev
18:42 shaunm_ joined #gluster-dev
19:11 aravindavk joined #gluster-dev
19:20 ashiq- joined #gluster-dev
20:21 kaushal_ joined #gluster-dev
20:25 shyam left #gluster-dev
20:46 glusterbot joined #gluster-dev
20:50 glusterbot joined #gluster-dev
20:55 glusterbot` joined #gluster-dev
20:55 glusterbot joined #gluster-dev
20:56 glusterbot joined #gluster-dev
20:56 glusterbot_ joined #gluster-dev
20:57 glusterbot joined #gluster-dev
20:58 glusterbot joined #gluster-dev
20:58 shaunm_ joined #gluster-dev
20:58 glusterb8t joined #gluster-dev
20:59 glusterbot joined #gluster-dev
21:20 badone joined #gluster-dev
21:39 shyam joined #gluster-dev
21:39 shyam left #gluster-dev
23:02 wushudoin| joined #gluster-dev
23:07 shyam joined #gluster-dev
23:07 wushudoin| joined #gluster-dev
23:13 wushudoin| joined #gluster-dev
23:13 _Bryan_ joined #gluster-dev
23:14 ira joined #gluster-dev

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