Camelia, the Perl 6 bug

IRC log for #gluster-dev, 2013-05-27

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

All times shown according to UTC.

Time Nick Message
01:27 jclift_ joined #gluster-dev
01:29 jclift joined #gluster-dev
02:56 lalatenduM joined #gluster-dev
03:10 bharata joined #gluster-dev
03:22 shubhendu joined #gluster-dev
03:45 mohankumar__ joined #gluster-dev
03:53 _BryanHm_ joined #gluster-dev
04:15 bulde joined #gluster-dev
04:50 aravindavk joined #gluster-dev
05:08 hagarth joined #gluster-dev
05:18 lalatenduM joined #gluster-dev
05:21 lala_ joined #gluster-dev
05:28 rastar joined #gluster-dev
06:03 rgustafs joined #gluster-dev
06:11 bala1 joined #gluster-dev
06:24 vshankar joined #gluster-dev
06:46 kaushal joined #gluster-dev
06:56 mohankumar__ hagarth: is krishnan parthasarathi available in irc?
07:04 hagarth mohankumar__: will ask him to join in
07:04 mohankumar__ hagarth: thanks
07:13 mohankumar__ hagarth: are you ok with the idea of restricting --remote-host usecase?
07:36 puebele joined #gluster-dev
07:44 shubhendu joined #gluster-dev
07:54 puebele joined #gluster-dev
08:21 krishnan_p joined #gluster-dev
08:59 bharata joined #gluster-dev
09:05 kaushal_ joined #gluster-dev
09:19 tziOm joined #gluster-dev
09:20 tziOm Playing around with https://github.com/avati/samba.git I get the following error when trying to connect to a gluster vfs share:
09:20 tziOm ./sbin/smbd: symbol lookup error: /opt/samba/lib/vfs/glusterfs.so: undefined symbol: glfs_new
09:24 hagarth joined #gluster-dev
09:29 susant1 joined #gluster-dev
09:32 rastar joined #gluster-dev
09:33 susant1 left #gluster-dev
09:46 _ndevos tziOm: do you have libgfapi.so in the 'ldconfig -p'?
09:47 ndevos tziOm: and, does 'ldd /opt/samba/lib/vfs/glusterfs.so' show that libgfapi is resolved correctly?
09:53 tziOm ndevos, it was not linked in, so did it manually and it worked
09:53 kshlm joined #gluster-dev
09:53 tziOm samba shows "LIBFLAGS" correctly, but does not link glusterfs.so with -lgfapi
09:54 tziOm probably small bug in makefile
09:54 tziOm Anyway.. now it is kind of working...
09:54 tziOm Successfully loaded vfs module [glusterfs] with the new modules system
09:54 ndevos tziOm: yes, sounds like it
09:55 ndevos tziOm: very cool!
09:55 tziOm but:
09:55 tziOm make_connection: VFS make connection failed!
09:55 * ndevos doesnt know much about that
09:55 tziOm is this module supposed to work at all..?
10:01 mohankumar__ krishnan_p: ping
10:01 krishnan_p mohankumar__ pong
10:01 mohankumar__ apart from 127.0.0.1/32 you dont have any issue with the patch?
10:01 mohankumar__ krishnan_p: ^^
10:02 blues-man joined #gluster-dev
10:03 krishnan_p mohankumar__, apart from the comments on gerrit, I have no reservations on the RFC patch
10:08 mohankumar__ krishnan_p: for multiple glusterds on a single machine, you create multiple loopback interfaces?
10:12 edward1 joined #gluster-dev
10:18 bharata joined #gluster-dev
10:18 krishnan_p mohankumar__, yeah. we use 127.0.0.x, where x =1,2,...
10:24 mohankumar__ krishnan_p: thanks
10:35 krishnan_p mohankumar__, cool
10:42 hagarth joined #gluster-dev
11:02 mohankumar__ krishnan_p: instead of denying the request at each gluster cli entry point (such as __glusterd_handle_create_volume)
11:02 mohankumar__ i want to do that when glusterd receives the request, where we can check the 'op' of CLI operation and deny the request
11:03 mohankumar__ thats better right?
11:04 jclift tziOm: As a thought, when you compiled that version of samba, what gluster-devel package did you have installed at the same time?
11:05 jclift tziOm: Was it a new gluster-devel package set build from git, or was it a gluster-devel package set provided by your distro?
11:05 jclift tziOm: Asking because I _think_ it needs to be a very very recent version compiled from git (like today or yesterday's git onwards).
11:06 * jclift Isn't totally sure though.  Wouldn't bet money on it, but worth trying just in case.
11:06 jclift :D
11:07 tziOm jclift, git today
11:07 jclift Oh well
11:07 tziOm so is it only me trying/interested in this?
11:08 jclift tziOm: It sounds like the best person to ping will be avati directly, as he's the coder for it. :)
11:09 jclift tziOm: Not sure who else is trying it out.  People being spread across timezone's, mailing list vs irc, etc.
11:10 jclift avati: You around?  If so, tziOm is trying out your samba stuff and it's breaking (so far)... able to help?
11:12 jclift tziOm: If you don't get any joy here on IRC, definitely communicate this stuff via gluster-devel.  Avati has been following up discussion about this stuff via the dev mailing list through the weekend.
11:20 tziOm avati: ?
11:39 krishnan_p mohankumar__, rpcsvc.c has rpc_handle_rpc_call which looksup the appropriate actor (handler) based on prog,procnum pair
11:40 krishnan_p mohankumar__,  ^^ s/looksup/looks up/
11:40 mohankumar__ krishnan_p: i updated gerrit regarding that
11:41 krishnan_p mohankumar__, But that would mean, you need to special case on glusterd's cli program. Which doesn't look good. or does it?
11:42 mohankumar__ krishnan_p: inside rpcsvc_handle_rpc_call(), based on req->procnum we can decide if its read-only operation or not
11:42 mohankumar__ and deny the request if the request came from remote-host
11:43 krishnan_p mohankumar__, But we want that behaviour to be restricted to the Gluster CLI program.
11:45 mohankumar__ krishnan_p: so procnum is not unique?
11:45 mohankumar__ i mean depending on actor they are unique?
11:46 krishnan_p Its not the uniqueness that concerns me. Is rpcsvc the right place to handle such program specific special-case'ing?
11:47 mohankumar__ that avoids having calling glusterd_rpc_deny_request() function in each gluster CLI proc handling functions
11:49 krishnan_p rpcsvc is expected to route inbound messages to appropriate actor based on prog, procnum.
11:49 krishnan_p rpcsvc also provide auth-allow functionality to treat (rpc) callers differently based on their network address
11:50 mohankumar__ krishnan_p: so there is no single actor specific entry point right?
11:50 krishnan_p What you want here is rpcsvc to use the knowledge that is specific to a higher level consumer (glusterd CLI prog) in this case, to deny request
11:51 krishnan_p to me, despite the tediousness involved in special case'ing for every cli handler, rpcsvc doesn't seem the right place
11:51 krishnan_p mohankumar__, at least, I don't know of any from the top of my mind
11:52 mohankumar__ krishnan_p: so you think current approach is acceptable?
11:55 krishnan_p mohankumar__, Yes, I do, until one of us comes up with a different one.
11:58 mohankumar__ krishnan_p: when i do a create volume from peer, the path is different ie __glusterd_handle_create_volume() is not called
11:58 mohankumar__ any idea how its done
12:19 tziOm Is it possible to have some caching with glusterfs?
12:23 krishnan_p joined #gluster-dev
12:51 krishnan_p joined #gluster-dev
13:07 puebele3 joined #gluster-dev
13:17 16WAADQW1 joined #gluster-dev
13:37 puebele1 joined #gluster-dev
13:56 puebele1 joined #gluster-dev
14:00 wushudoin joined #gluster-dev
14:01 mohankumar__ joined #gluster-dev
15:42 hagarth joined #gluster-dev
16:17 awheeler_ joined #gluster-dev
18:00 _BryanHm_ left #gluster-dev
22:09 yinyin joined #gluster-dev
23:24 yinyin joined #gluster-dev

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