Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2014-10-31

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

All times shown according to UTC.

Time Nick Message
02:21 kshlm joined #gluster-dev
02:26 anoopcs joined #gluster-dev
02:27 bala joined #gluster-dev
02:29 anoopcs joined #gluster-dev
02:56 kshlm joined #gluster-dev
03:32 kshlm joined #gluster-dev
03:33 jiffin joined #gluster-dev
03:35 shubhendu joined #gluster-dev
03:38 bala joined #gluster-dev
03:43 pranithk joined #gluster-dev
03:44 kanagaraj joined #gluster-dev
03:49 pranithk left #gluster-dev
03:51 hagarth joined #gluster-dev
04:03 bala joined #gluster-dev
04:08 jiffin joined #gluster-dev
04:28 ppai joined #gluster-dev
04:29 atinmu joined #gluster-dev
04:37 rafi1 joined #gluster-dev
04:49 nishanth joined #gluster-dev
05:00 itisravi joined #gluster-dev
05:01 lalatenduM joined #gluster-dev
05:04 Gaurav_ joined #gluster-dev
05:05 kshlm joined #gluster-dev
05:09 hagarth joined #gluster-dev
05:11 atalur joined #gluster-dev
05:14 ndarshan joined #gluster-dev
05:23 kshlm joined #gluster-dev
05:31 kdhananjay joined #gluster-dev
05:31 kdhananjay left #gluster-dev
05:58 deepakcs joined #gluster-dev
06:03 aravindavk joined #gluster-dev
06:09 ppai joined #gluster-dev
06:22 pranithk joined #gluster-dev
06:34 hchiramm_ joined #gluster-dev
06:56 shubhendu joined #gluster-dev
06:58 rgustafs joined #gluster-dev
07:14 soumya joined #gluster-dev
07:17 atinmu joined #gluster-dev
07:18 aravindavk joined #gluster-dev
07:24 shubhendu joined #gluster-dev
07:26 atalur joined #gluster-dev
07:31 ppai joined #gluster-dev
07:34 bala joined #gluster-dev
07:36 rgustafs joined #gluster-dev
08:22 rgustafs joined #gluster-dev
08:40 badone_ joined #gluster-dev
08:42 badone__ joined #gluster-dev
08:49 vikumar joined #gluster-dev
09:02 ppai joined #gluster-dev
09:03 jiffin1 joined #gluster-dev
09:05 atalur joined #gluster-dev
09:20 shubhendu joined #gluster-dev
09:20 soumya joined #gluster-dev
09:20 ndarshan joined #gluster-dev
09:21 Guest74799 joined #gluster-dev
09:24 atinmu joined #gluster-dev
09:26 bala joined #gluster-dev
09:46 ndevos lalatenduM, Humble: how do you think about a 3.5.3 beta2 now?
09:47 bala joined #gluster-dev
09:55 lalatenduM ndevos, yeah it is fine , we can do that. but not on the same day of 3.6 GA , it will not get much attention :)
09:57 ndevos lalatenduM: its building now, and you should get an email when its ready :)
09:57 hagarth kshlm: http://blog.gluster.org/tag/block/
09:58 ndevos lalatenduM: its up to you when you want to build the rpms, I can do a blog post when those are available
10:06 jiffin joined #gluster-dev
10:06 lalatenduM ndevos, got the mail
10:06 lalatenduM :)
10:08 lalatenduM ndevos, sure will work on it
10:10 shubhendu joined #gluster-dev
10:11 ndevos lalatenduM++ thanks
10:11 glusterbot ndevos: lalatenduM's karma is now 31
10:14 atinmu joined #gluster-dev
10:15 ndarshan joined #gluster-dev
10:23 kdhananjay joined #gluster-dev
10:24 soumya joined #gluster-dev
10:27 shubhendu joined #gluster-dev
10:28 hagarth ndevos++ & xavih++ - incorporated your review comments
10:28 glusterbot hagarth: ndevos's karma is now 40
10:28 glusterbot hagarth: xavih's karma is now 6
10:28 hagarth for release notes i.e.
10:29 xavih hagarth: can you look at a mail I've recently sent to gluster-devel ?
10:29 hagarth xavih: have seen that.. how do you think we can fix it?
10:29 xavih hagarth: I think there's a dangerous bug in memory pool framework
10:30 xavih I think the problem is the definition of GF_MEM_POOL_PAD_BOUNDARY
10:30 xavih hagarth: it uses sizeof(int), which is 4
10:31 xavih hagarth: a quick fix would be to use sizeof(uintptr_t) to use the natural size of a pointer the architecture where it's compiled
10:31 xavih hagarth: it will use some more memory though
10:32 hagarth xavih: yeah, seems like the right thing to me
10:32 xavih hagarth: I think that 8 bytes alignment is enough to avoid this problem
10:32 xavih hagarth: Do I file a bug a send a patch ?
10:33 hagarth xavih: go ahead
10:34 xavih hagarth: ok
10:35 kshlm hagarth, Humble, https://github.com/docker/docker/issues/8881
10:35 kshlm Isn't neependra here?
10:35 kshlm hagarth, could you review the issue?
10:38 hagarth kshlm: will do
10:38 hagarth kshlm: looks good
10:42 Guest74799 joined #gluster-dev
10:46 ndevos kshlm: does that requite "FUSE mounts from user namespaces"? see bottom of http://www.linuxplumbersconf.org/2014/wp-content/uploads/2014/10/LPC2014_Containers.txt
10:47 ndevos kshlm: from what I understood, mounting a fuse filesystem in a container is not very well supported/secure yet
10:50 bala joined #gluster-dev
10:55 Humble kshlm++ looks good
10:55 glusterbot Humble: kshlm's karma is now 3
10:56 kshlm ndevos, I really don't know of the requirements needed to make it happen.
10:56 kshlm But the use case of having glusterfs providing docker volumes is something that is nice to have.
10:57 ndevos kshlm: oh, yes, it would be nice indeed!
11:00 nkhare joined #gluster-dev
11:03 shubhendu joined #gluster-dev
11:25 lalatenduM_ joined #gluster-dev
11:27 shubhendu joined #gluster-dev
11:42 soumya joined #gluster-dev
11:46 soumya_ kkeithley, ndevos ,
11:46 soumya_ ping
11:58 hagarth1 joined #gluster-dev
12:01 jdarcy joined #gluster-dev
12:09 edward1 joined #gluster-dev
12:14 soumya_ joined #gluster-dev
12:26 pranithk left #gluster-dev
12:42 ws2k3 will glusterfs 3.6 have official freebsd support ?
12:53 kkeithley I'm not really sure what "official freebsd support" means. It builds and runs on freebsd. There is a maintainer from the FreeBSD community who is packaging it (as a port?) for FreeBSD. The intent is to preserve that. But like a lot of other things, e.g. NetBSD and Mac OS X support, these things come from the community. E.g. Solaris support has withered on the vine because, apparently, the community doesn't value it enough to do the w
12:55 kkeithley It's all community. There is no official or unofficial. If the community decides to do it, it happens.
12:56 bala joined #gluster-dev
13:08 ws2k3 kkeithley is the build for freebsd stable ?
13:09 ws2k3 that is what i mean with official
13:09 ndevos soumya++
13:09 glusterbot ndevos: soumya's karma is now 2
13:15 ws2k3 Harshavardhana is the freebsd port maintainer right ? is he from glusterfs or is that someone from freebsd ?
13:20 JustinClift ws2k3: Harsha is the primary coder making the port happen atm
13:20 JustinClift ws2k3: Craig Butler is in process to become the port maintainer though
13:20 kkeithley Is it stable? Yes, I think it is
13:22 JustinClift ws2k3: Um, maybe email Craig Butler and ask him how far off the port is looking
13:32 shyam joined #gluster-dev
13:36 kkeithley packages for 3.6.0 will be available shortly?
13:40 ndevos kkeithley: lol, I was wondering about that too :D
13:42 Humble lalatenduM++
13:42 glusterbot Humble: lalatenduM's karma is now 32
13:43 Humble kkeithley, we will skip 3.6.0 packages
13:43 lalatenduM_ :)
13:44 Humble kkeithley, ndevos lalatenduM_ any objection  ? :P
13:44 rgustafs joined #gluster-dev
13:45 Humble hopefully we will have 3.6.1 in early next week
13:45 Humble so we will start with that
13:45 lalatenduM_ Humble,  we need official word on 3.6.1 then
13:46 ndevos Humble: I really don't care much, but I was wondering about the denial of 3.6.0 packages that was in Kalebs (?) email
13:46 Humble ndevos, I think u were away when we were discussing about this decision ..  :)
13:46 ndevos and, does that mean we need to explain the yum-priorites solution for RHEL/CentOS installations too?
13:47 lalatenduM_ I mean we need to send the mail to community mailing lists abt out our decision not to have RPMs for 3.6.0 and plan for 3.6.1
13:49 lalatenduM_ ndevos, yes, we need to explain the issue caused by 3.6.29-2 gluster client packages in RHEL 6.6
13:49 ndevos lalatenduM_: hagarth mentions that packages for 3.6.0 will be made available soon?
13:50 kkeithley no objection. I thought we (where we was the set of Humble, Lala, hagarth, ndevos, and me, with nobody else weighing in) had agreed we would not package 3.6.0.
13:50 kkeithley ndevos: yes, hence my question, given what I thought "we" decided here yesterday
13:51 Humble ndevos, kkeithley lalatenduM_ the decision is no packages for 3.6.0
13:51 Humble but I see there is a mention in hagarth's mail.
13:51 kkeithley Humble: yes, that was my understanding.
13:51 kkeithley indeed, hence my question
13:51 Humble so 2 options for us.
13:52 Humble either we can notify the community that packages will only be available for 3.6.1 which is going to happen soon.
13:52 Humble or if some one requested the package , we can respond to the thread
13:52 Humble but again the decision is no packages for 3.6.0 ..
13:52 ndevos I'd say wait for 3.6.1, and reply to hagarths email now
13:53 Humble yep ..
13:53 kkeithley +1
13:53 Humble in other view,
13:53 Humble if some one semiosis want to build packages for 3.6.0 he should be free to do it
13:54 Humble so that the justification for the strings "Packages for various distributions will be available shortly at the download site." comes in
13:54 ndevos yes, indeed, we dont do 3.6.0 for RHEL/CentOS, other distributions (Debian, Ubuntu, Fedora) can have 3.6 packages
13:54 Humble yep .. thats what the right though I believe.
13:54 lalatenduM_ ndevos, agree
13:54 Humble semiosis, ^^^ :)
13:55 ndevos who's sending the email?
13:56 Humble ndevos, we need to get confirmation from semiosis before sending that email
13:57 Humble in downl.g.o the fedora/centos/rhel folders can have a text file with some strings..
13:57 Humble lalatenduM++ is making it :)
13:57 glusterbot Humble: lalatenduM's karma is now 33
13:57 ndevos Humble: we can provide 3.6.0 for Fedora, that is not an issue
13:57 kkeithley if we package for Fedora, with libgfapi.so.7.0.0, and then package 3.6.1 with libgfapi.so.0.0.0+symbolversions, what happens?
13:57 kkeithley more breakage?
13:58 ndevos I'm not sure symbolversions can be used as a dependency in rpm
13:58 lalatenduM_ kkeithley, right
13:58 Humble yeah.. it can introduce more breakage
13:59 ndevos we can only rely on symbolversions when new versions from qemu etc. have a requires on the libgfapi.so.*...something...SYMBOL(...) or whatever
14:00 ndevos binaries do that, but does rpm have that capability too?
14:00 * lalatenduM_ is not sure
14:01 kkeithley all true I suppose. But first, if we do 3.6.0 packages for Fedora, we have to send email to fedora-devel notifying of the SO_NAME bump
14:02 lalatenduM_ kkeithley, right
14:02 kkeithley and if we go back to libgfapi.so.0.0.0 in 3.6.1 we'll have to notify again, and we'll need a hack compat symlink and....
14:02 Humble I still think there is no need to create packages for fedora.
14:02 lalatenduM_ kkeithley, +1
14:02 kkeithley not if we're going to do 3.6.1 in a couple days
14:02 Humble kkeithley, 3.6.1 is happening in next week
14:03 Humble all the afrv2 patches are already available in 3.6.0 build..
14:03 deepakcs joined #gluster-dev
14:04 kkeithley the main reason for 3.6.1 is to mitigate the 3.6.0 mess from RHEL and RHS
14:04 Humble indeed .. thats the aim
14:04 kkeithley with or without any changes
14:04 Humble yep
14:04 kkeithley we could just tag 3.6.0 with a 3.6.1 tag and release again
14:04 ndevos just update the release notes :D
14:04 kkeithley but since we have a couple days we can possibly address the libgfapi.so.X issues
14:05 Humble yeah. if in case some patches came in , we can think those as well ..
14:05 kkeithley and maybe a couple small patches hagarth has
14:05 Humble yep
14:05 kkeithley sorry, I know everyone knows these things
14:05 Humble :) .. thanks for double confirmation :)
14:05 lalatenduM_ kkeithley, np
14:06 kkeithley did I miss anything?
14:06 ndevos I think we still need the SO_NAME bump from .so.0 -> .so.7, with or without symbol versioning
14:07 ndevos I just do not think rpm does the symbolversioning in RHEL/CentOS, maybe it is a new feature available in Fedora
14:07 kkeithley I'm not aware of anything in RPM that knows about symbol versioning
14:07 JustinClift Scanning over ^^^, we're thinking of doing a 3.6.1 in a few days?
14:08 Humble JustinClift, yep ..
14:09 ndevos hmm, https://www.berrange.com/posts/2011/01/13/versioning-in-the-libvirt-library/ seems to use symbol versioning, but it looks like versioning without a relation to the symbol
14:11 kkeithley libc has all kinds of versioned symbols
14:12 ndevos yes, but 'rpm -q --provides glibc' does not list the symbols, only the .so.X(SOME_VERSION)
14:12 ndevos but, that probably is sufficient?
14:13 lalatenduM_ Humble, kkeithley, devos , regarding the mail abt (rpms for rhel, fedora)  to community mailing lists. we will wait till Monday , if we don't release 3.6.1 on monday or tuesday, we will send mail abt it (plan for 3.6.1)
14:13 lalatenduM_ is that ok?
14:13 ndevos lalatenduM_: ok
14:13 Humble ack :)
14:13 kkeithley I doubt I can have versioned symbols for libgfapi by Monday. At least not through review
14:17 kkeithley right now I'm compiling a matrix of which functions appeared in which versions.
14:18 ndevos kkeithley: do let us know if you need any assistance!
14:18 kkeithley nearly done. I'll post link to google drive spreadsheet in a few mintues
14:19 ndevos google drive?!
14:19 kkeithley needed a spreadsheet to easily insert columns.
14:19 kkeithley it's community
14:19 kkeithley ;-)
14:19 ndevos yeah, sure
14:20 kkeithley don't drink the KoolAid
14:20 ndevos ethercalc!
14:20 kkeithley too many choices
14:20 ndevos :P
14:21 kkeithley lol
14:24 lalatenduM_ kkeithley, we can again review the situation on monday , decide accordingly
14:25 kkeithley choices I don't necessarily even know about until after the fact
14:25 kkeithley lalatenduM_: sure
14:45 kkeithley ethercalc, just for ndevos. ;-)   https://ethercalc.org/lrjvqrapzu
14:46 kkeithley matrix of the public gfapi APIs, and the version they appeared in
14:48 kkeithley need to decide what the symbol versions would be, accordingly.
14:48 kkeithley And how did we arrive at 7 for the current libgfapi.so?
14:49 ndevos kkeithley++ :D
14:49 glusterbot ndevos: kkeithley's karma is now 30
14:50 kkeithley cut-and-paste FTW
14:50 ndevos more symbols than I thought!
14:52 kkeithley 1.0 and 1.1 for 3.4.0 and 3.4.2 respectively.
14:52 kkeithley 2.0 and 2.1 for 3.5.0 and 3.5.1.
14:52 kkeithley 3.0 for 3.6.0?
14:54 ndevos why not keep them versions aligned?
14:55 kkeithley yeah, could do that
14:56 ndevos I'd prefer that, it makes it easier to see whats wrong in case rpm can not resolve something
14:58 shubhendu joined #gluster-dev
15:00 ndevos kkeithley: hmm, do we know what SO_NAME in RHS glusterfs-3.6 is used? maybe we do need the .so.7 for RHEL/CentOS
15:00 kkeithley good question. I wondered about that.
15:00 * ndevos checks
15:02 ndevos kkeithley: we're lucky there, libgfapi.so.0 :D
15:02 kkeithley yup, glusterfs-api-3.6.0.28-1.el6rhs.x86_64.rpm -> /usr/lib64/libgfapi.so.0.0.
15:02 kkeithley RHS 3.0/3.0-GOLD
15:03 kkeithley is that what you looked at?
15:03 ndevos I checked 3.6.0.30
15:03 kkeithley kewl....
15:04 kkeithley oh yeah, bala sends out email with the newest build location
15:05 ndevos oh, I just download the latest package from $server
15:06 * lalatenduM_ is wondering if this is #rhs or #gluster-dev  channel :)
15:16 kkeithley and yeah, client-side bits are in RHEL
15:16 kkeithley are in RHEL too
15:39 * ndevos still needs to have lunch... I'll be back later
15:46 * kkeithley needs lunch too, wonders if it isn't supper time in NL
16:01 lalatenduM_ kkeithley++ liked ur joke :), please dont go near any bus
16:01 glusterbot lalatenduM_: kkeithley's karma is now 31
16:01 semiosis will build 3.6.0 today
16:04 hagarth joined #gluster-dev
16:07 ndevos semiosis: just be careful with it, it has libgfapi.so.7 and that would require rebuilds of related packages
16:08 ndevos semiosis: that is an issue we do not like too much, and we may go back to .so.0 and add additional symbol versioning instead
16:09 semiosis ndevos: thx for the heads up.
16:10 ndevos semiosis: we like to see that 'fixed' in 3.6.1, which hopefully gets released early next week
16:11 ndevos the plan is to not provide RPMs for 3.6.0 for now...
16:13 semiosis oh, well then
16:13 semiosis i can wait until 3.6.1 to do debs
16:31 jobewan joined #gluster-dev
16:34 davemc GlusterFS 3.6.0 is alive! Just in time for your holiday treat, bluster community has released 3.6.0. More at http://blog.gluster.org/?p=10661
16:34 semiosis damn autocorrect
16:35 davemc GlusterFS 3.6.0 is alive! Just in time for your holiday treat, gluster community has released 3.6.0. More at http://blog.gluster.org/?p=10661
16:38 davemc hagarth, did announce email go out? My mail system is unhappy with me today
16:39 hagarth davemc: yes, have sent a note to both users and devel MLs
16:39 davemc hagarth, should it go to announce and board as well?
16:39 hagarth davemc: we could send it over there
16:40 davemc as soon as it its my inbox
19:16 lalatenduM joined #gluster-dev
20:16 shyam joined #gluster-dev
23:25 pranithk joined #gluster-dev

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