Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2014-01-07

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

All times shown according to UTC.

Time Nick Message
00:04 sac`away joined #gluster-dev
01:04 yinyin joined #gluster-dev
03:06 kshlm joined #gluster-dev
03:09 bharata-rao joined #gluster-dev
03:22 shubhendu joined #gluster-dev
03:40 badone__ joined #gluster-dev
03:47 itisravi joined #gluster-dev
04:06 bala joined #gluster-dev
04:36 shyam joined #gluster-dev
04:45 shubhendu joined #gluster-dev
04:45 spandit joined #gluster-dev
04:56 hagarth joined #gluster-dev
04:57 ndarshan joined #gluster-dev
04:59 mohankumar joined #gluster-dev
05:00 ababu joined #gluster-dev
05:01 lalatenduM joined #gluster-dev
05:11 kdhananjay joined #gluster-dev
05:12 ppai joined #gluster-dev
05:25 yinyin joined #gluster-dev
05:29 mohankumar joined #gluster-dev
05:31 ndarshan joined #gluster-dev
05:32 nixpanic_ joined #gluster-dev
05:32 nixpanic_ joined #gluster-dev
05:36 aravindavk joined #gluster-dev
05:36 bharata-rao joined #gluster-dev
05:36 lalatenduM joined #gluster-dev
05:41 inodb_ joined #gluster-dev
05:41 ajha joined #gluster-dev
05:49 nixpanic joined #gluster-dev
05:50 nixpanic joined #gluster-dev
06:04 raghu joined #gluster-dev
06:09 bala joined #gluster-dev
08:20 bala joined #gluster-dev
08:51 badone__ joined #gluster-dev
08:55 hagarth joined #gluster-dev
09:06 badone__ joined #gluster-dev
09:13 yinyin joined #gluster-dev
09:23 shubhendu joined #gluster-dev
09:45 badone joined #gluster-dev
09:58 jclift joined #gluster-dev
10:37 jclift left #gluster-dev
10:41 jclift joined #gluster-dev
10:52 shubhendu joined #gluster-dev
11:02 kanagaraj joined #gluster-dev
11:16 kshlm joined #gluster-dev
11:18 shubhendu joined #gluster-dev
11:19 spandit joined #gluster-dev
11:32 kaushal_ joined #gluster-dev
11:41 kkeithley1 joined #gluster-dev
11:43 jclift Before I went on holiday, someone mentioned there's a new guy (intern maybe?) who has been tasked with improving libgfapi.
11:43 jclift Does anyone remember who that new guy is?
11:43 * jclift can't find it in email, and doesn't remember
11:54 shyam joined #gluster-dev
12:06 shubhendu joined #gluster-dev
12:07 spandit joined #gluster-dev
12:07 itisravi joined #gluster-dev
12:11 foster ndevos|mtg: around?
12:12 ndevos|mtg foster: yes, just, almost having lunch
12:12 foster just curious about your gluster mt mail... did write-behind play any factor in that scenario?
12:12 foster (glusterfs causes multi-threaded behavior ...)
12:13 ndevos foster: yeah, seems like it - shyam pointed me in the same direction
12:14 ndevos foster: in this case, a manual ftruncate() (seek() + write()) is not detected correctlt, using ftruncate() prevents the unexpected behavior
12:14 ndevos s/correctlt/correctly/
12:15 foster by that you mean an extending truncate..?
12:15 ndevos yes, correct
12:16 foster interesting, ok. well my question was more wondering how the single threaded client would even get to mmap without a write completion
12:16 ndevos I've just started looking at write-behind's writev() implementation, and it does not consider EOF
12:16 foster which is more of a wb thing than an io-threads thing
12:17 ndevos I think write-behind/writev() should mark the - if offset > EOF - EOF+offset+size as dirty
12:18 ndevos well, something like that anyway, the math is incorrect
12:19 foster yeah. I'm not sure the posix rules.. but I would think even without i_size modifications a read after a write should return the previously written data
12:19 foster the i_size change is probably just what throws a wrench into things :P
12:19 ndevos yes, but i_size is checked in posix_readv()...
12:20 foster is this on a fuse client btw?
12:20 ndevos yes, write-behind on a fuse client, removing write-behind from the graph prevents the problem
12:21 ndevos but, I'll be having luch now, will be back later
12:21 foster ok
12:21 kkeithhley_ according to git blame, Shyam wrote the new gfapi handles support for ganesha. Is that who you were trying to remember. I see other commits in gfapi by Poornima.
12:21 kkeithhley_ jclift: ^^^
12:30 ababu joined #gluster-dev
12:34 jclift kkeithhley_: Hmmmm, not sure.  I'll ping both of them I guess and find out. :D
12:35 jclift kkeithhley_: Thx btw. :)
12:55 hagarth joined #gluster-dev
13:10 kaushal_ joined #gluster-dev
13:49 JoeJulian Is there a way to 1) retrieve the fuse volfile through gfapi, or do I need to use libglusterfs; 2) use a volfile instead of a volfile-server in gfapi?
13:58 lalatenduM joined #gluster-dev
14:00 ira joined #gluster-dev
14:01 ira joined #gluster-dev
14:06 edward2 joined #gluster-dev
14:06 kanagaraj joined #gluster-dev
14:36 JoeJulian Ah, 2 is get_volfp
14:49 JoeJulian ... or not...
14:52 jclift ndevos: Any idea about the answer to JoeJulian's question ^ ?
14:53 JoeJulian Ah, ok, just don't specify a server and use the vol file as the volume name.
14:53 jclift ndevos: nvm :D
14:53 ndevos :)
14:54 JoeJulian Still want to know how to retrieve the fuse volfile initially though. I have a cool idea for split-brain administration.
14:54 JoeJulian I did a poc by hand yesterday and it worked elegantly, and it's simple.
14:58 ndevos the GETSPEC procedure returns the volfile iirc, I'm not sure if that is exposed in libgfapi
15:03 ndevos JoeJulian: I think you can access (struct glfs*)->oldvolfile (it's the contents of the volfile) and ->oldvollen for the size of ->oldvolfile
15:04 ndevos well, you'll need to include glfs-internal.h to get to those attributes :-/
15:05 JoeJulian I'm doing this in python, but I can define that structure. Cool.
15:05 ira_ joined #gluster-dev
15:06 hagarth JoeJulian: gluster system:: getspec <volname>
15:08 JoeJulian hagarth: I'd prefer to do this from a non-server.
15:09 hagarth JoeJulian: --remote might work for this read-only command
15:11 kkeithley1 joined #gluster-dev
15:32 wushudoin joined #gluster-dev
16:11 hagarth joined #gluster-dev
16:14 bala joined #gluster-dev
16:35 kanagaraj joined #gluster-dev
16:50 aravindavk joined #gluster-dev
16:55 jclift left #gluster-dev
17:02 jclift joined #gluster-dev
17:10 Technicool joined #gluster-dev
17:34 semiosis johnmark: just published glusterfs 3.4.2 packages for debian wheezy on download.gluster.org
17:35 jobewan joined #gluster-dev
18:01 jclift left #gluster-dev
18:04 johnmark semiosis: wooo
18:40 wushudoin joined #gluster-dev
18:41 avati joined #gluster-dev
19:01 a2_ joined #gluster-dev
19:33 * johnmark shakes fist at Middleman
21:13 ira joined #gluster-dev
21:21 _Bryan_ joined #gluster-dev
22:45 primechu_ joined #gluster-dev
23:08 ira joined #gluster-dev

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