Perl 6 - the future is here, just unevenly distributed

IRC log for #gluster-dev, 2015-06-15

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

All times shown according to UTC.

Time Nick Message
01:18 shyam joined #gluster-dev
01:48 ilbot3 joined #gluster-dev
01:48 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/
03:20 kdhananjay joined #gluster-dev
03:22 overclk joined #gluster-dev
03:56 shubhendu joined #gluster-dev
03:58 itisravi joined #gluster-dev
03:59 josferna joined #gluster-dev
04:15 nbalacha joined #gluster-dev
04:18 kshlm joined #gluster-dev
04:18 kshlm joined #gluster-dev
04:18 shaunm joined #gluster-dev
04:26 vimal joined #gluster-dev
04:33 sakshi joined #gluster-dev
04:36 atinm joined #gluster-dev
04:37 ppai joined #gluster-dev
04:42 pppp joined #gluster-dev
04:44 schandra joined #gluster-dev
04:47 nkhare joined #gluster-dev
04:54 nkhare joined #gluster-dev
04:58 ndarshan joined #gluster-dev
05:01 gem joined #gluster-dev
05:05 ashishpandey joined #gluster-dev
05:07 krishnan_p joined #gluster-dev
05:21 kdhananjay joined #gluster-dev
05:24 kdhananjay joined #gluster-dev
05:33 spandit joined #gluster-dev
05:34 deepakcs joined #gluster-dev
05:40 soumya joined #gluster-dev
05:42 overclk joined #gluster-dev
05:44 kdhananjay joined #gluster-dev
05:45 hagarth joined #gluster-dev
05:48 anrao joined #gluster-dev
05:49 josferna joined #gluster-dev
05:51 Gaurav__ joined #gluster-dev
05:55 hchiramm anrao, ping
05:55 glusterbot hchiramm: Please don't naked ping. http://blogs.gnome.org/mark​mc/2014/02/20/naked-pings/
05:55 atalur joined #gluster-dev
05:56 ashiq joined #gluster-dev
05:57 hchiramm anrao++
05:57 glusterbot hchiramm: anrao's karma is now 7
05:57 hchiramm ashiq, ping
05:57 glusterbot hchiramm: Please don't naked ping. http://blogs.gnome.org/mark​mc/2014/02/20/naked-pings/
05:57 Manikandan joined #gluster-dev
06:04 kdhananjay joined #gluster-dev
06:07 ashiq overclk, raghu : http://review.gluster.org/10297 is a bitrot logging patch please look into it :)
06:08 overclk sure ashiq. probably by today evening.
06:09 ashiq overclk : thanks :)
06:24 owlbot` joined #gluster-dev
06:25 pranithk joined #gluster-dev
06:30 rastar_afk joined #gluster-dev
06:32 hchiramm_ http://review.gluster.org/#/c/9836/ atinm
06:32 hchiramm_ can u please merge above patch
06:33 krishnan_p hchiramm, ndevos, hagarth, I see that github's markdown rendering of 3.7.1 release notes is screwed. This is due to incorrect use of unordered list syntax I had used. I have sent a fix for this - http://review.gluster.org/11219
06:33 hchiramm_ krishnan_p, reviewing it
06:33 krishnan_p hchiramm, thanks.
06:33 hchiramm_ atinm, that got +1 from sachin on friday and was waiting for votes from build system
06:33 hchiramm_ now we have the votes
06:35 hchiramm_ krishnan_p, I think it still need some more attention for rendering
06:35 hchiramm_ I was checking it in markable.in
06:35 krishnan_p hchiramm, OK. I used pandoc to convert into a single html file and saw that firefox renderred it fin.
06:36 krishnan_p hchiramm, pandoc -f markdown -t html doc/release-notes/3.7.1.md -o /tmp/3.7.1.html
06:36 hchiramm_ yeah , this conversion can be bit of failure sometimes.. :)
06:37 hchiramm_ I normally check it in markable.in or dilinger.io
06:38 hchiramm_ atinm, http://review.gluster.org/#/c/9864/ this as well
06:38 hchiramm_ now we have votes from build systems
06:38 hchiramm_ gem, as already addressed the comments
06:39 anrao joined #gluster-dev
06:42 krishnan_p hchiramm, why do you pandoc's conversion can fail at times
06:42 krishnan_p hchiramm, s/do you/do you say
06:43 hchiramm_ the format conversion use some rules which can fail
06:43 hchiramm_ it depends on input and output formats though
06:44 krishnan_p hchiramm, OK. Isn't markdown the input format and html the output format?
06:44 krishnan_p hchiramm, is there anything wrong with that?
06:45 hchiramm_ krishnan_p, I dont think anything wrong in html output..
06:46 hchiramm_ however the md itself not renderging properly in markable.in
06:46 hchiramm_ u can actually check it out , by pasting it in markable.in
06:46 krishnan_p hchiramm, OK. Why is that the litmus test? Is that similar to github's md rendering engine?
06:48 hchiramm_ the github rendering should be same ..  I do that test or write directly in markdown editors to make sure it will render properly
06:48 soumya joined #gluster-dev
06:54 nbalacha joined #gluster-dev
06:54 krishnan_p hchiramm, makes sense. I prefer having locally installed tools for pre-commit checks
06:54 hchiramm_ krishnan_p++ thanks
06:54 glusterbot hchiramm_: krishnan_p's karma is now 7
06:54 hchiramm_ atinm++ thanks !!
06:54 glusterbot hchiramm_: atinm's karma is now 3
06:55 spalai joined #gluster-dev
06:56 atinm hchiramm, done
06:56 hchiramm_ atinm, awesome!
06:56 hchiramm_ gem, ^^^
06:56 hchiramm_ gem, backport that to release 3.7 branch
06:56 gem hchiramm_, sure
06:56 hchiramm_ gem, thanks
06:57 hchiramm_ gem, I will retrigger netbsd tests for rest of 3 patches.
06:58 hchiramm_ u have addressed all the comments on review tright ?
06:58 gem hchiramm_, yes, I have
06:58 hchiramm_ k
07:01 hchiramm_ gem, I have retriggered netbsd on http://review.gluster.org/#/c/10473/ http://review.gluster.org/#/c/9905/ http://review.gluster.org/#/c/9877/
07:01 hchiramm_ all three have +1 from review and gluster build system vote
07:02 gem hchiramm_, thanks!
07:02 hchiramm_ so as soon as netbsd vote u got, please ping atinm or glusterd folks to merge it
07:02 gem hchiramm_++
07:02 glusterbot gem: hchiramm_'s karma is now 11
07:02 hchiramm_ its important
07:02 gem hchiramm_, yes!
07:02 hchiramm_ thanks
07:04 hchiramm_ anrao, ping
07:04 glusterbot hchiramm_: Please don't naked ping. http://blogs.gnome.org/mark​mc/2014/02/20/naked-pings/
07:05 hchiramm_ have u addressed nithya's comments in patch set v15 ?
07:05 sakshi joined #gluster-dev
07:08 soumya joined #gluster-dev
07:15 rjoseph joined #gluster-dev
07:18 anrao joined #gluster-dev
07:18 hchiramm_ anrao, ping
07:18 glusterbot hchiramm_: Please don't naked ping. http://blogs.gnome.org/mark​mc/2014/02/20/naked-pings/
07:24 ndevos FYI: in case you see NetBSD regressions hanging on "mount_nfs: can't access /patchy: Permission denied" - reboot the VM, retrigger the regression test
07:25 ndevos it most likely is caused by a crashed gluster/nfs server, check: gdb --core=/glusterfsd.core /buiuld/install/sbin/glusterfsd
07:27 ndevos auth_cache_lookup() should use refcounting for the lookup_res->item structure, patch is getting re-posted soon
07:39 anrao joined #gluster-dev
08:07 anekkunt joined #gluster-dev
08:09 shubhendu joined #gluster-dev
08:18 nbalacha joined #gluster-dev
08:25 anrao joined #gluster-dev
08:35 rgustafs joined #gluster-dev
08:43 gem joined #gluster-dev
08:49 ndarshan joined #gluster-dev
08:57 RaSTar any reviews for https://github.com/gluster/gluste​rfs-patch-acceptance-tests/pulls
08:58 RaSTar ^^ Had sent a patch to enable FORTIFY for regression builds
09:01 saurabh_ joined #gluster-dev
09:02 atinm RaSTar, O2 is must?
09:03 RaSTar atinm: yes, unfortunately
09:03 RaSTar so it is a choice between fortify and easy debugging of other cores
09:04 atinm RaSTar, will it cause any issue in terms of debugging?
09:05 RaSTar atinm: Debugging from cores should not change much. Some of the values might be optimized
09:06 RaSTar Hence the need for reviews, I am not sure of what all will change
09:06 RaSTar However, we can always try this for a week...
09:06 atinm RaSTar, we we make rpms, do we override these flags or we use the same mentioned in buid.sh ?
09:06 atinm *build.sh
09:07 soumya ndevos, is that the issue reported in 1230612?
09:07 RaSTar atinm: we make RPMs as per Fedora standard, it includes these flags by default
09:07 RaSTar atinm: so RPMs are always O2 and FORTIFY=2
09:08 atinm RaSTar, then we should be good :)
09:09 RaSTar atinm++ Thanks :) a comment on the pull request please
09:09 glusterbot RaSTar: atinm's karma is now 4
09:09 atinm RaSTar, already done
09:09 atinm :)
09:10 RaSTar ndevos: ^^ ,  Can we try the build.sh patch for a week?
09:10 ndevos RaSTar: I'm not sure about the requirement for -O2? it is not clear from "man 7 feature_test_macros"
09:12 ndevos soumya: that might be related, not sure... I would like to run that test with a debug build to see if it has more clear diagnostics
09:13 RaSTar ndevos:
09:13 RaSTar If _FORTIFY_SOURCE is set to 1, with compiler optimization level 1 (gcc -O1) and above, checks  that
09:13 RaSTar shouldn't  change  the behavior of conforming programs are performed.
09:13 soumya ndevos, okay
09:13 RaSTar We don't necessarily need O2, but O1 is a must. O0 that we use now, won't work.
09:13 ndevos RaSTar: yes, but I do not see how -O2 would be required for FORTIFY=2
09:14 RaSTar ndevos: Yes , I went with O2 as that is what we use with RPMs. Just to be uniform.
09:14 ndevos RaSTar: I do not know how much difference we would see in a core with -O0/1/2...
09:14 RaSTar ndevos: me too :(
09:14 ndevos RaSTar: well, debug/test builds are different from production builds, they would not need -O2
09:15 * ndevos is a little conservative
09:15 RaSTar ndevos:  yes, yes. But sometimes we may see some optimizations from O2 leading to errors.
09:16 RaSTar ndevos: like the infamous syncop bug that pranithk fixed
09:16 ndevos RaSTar: and I guess that we sometimes need -O0 to analyze other issues?
09:17 RaSTar ndevos: yes.. I am afraid of the cases where our regression will fail and we will have to re-compile with O0 and reproduce the issue.
09:18 RaSTar ndevos: but I don't see any other way. If we don't catch these in regression they will come from community users..
09:20 RaSTar ndevos: we can wait to get a good opinion.. from kkeithley maybe?...
09:20 ndevos RaSTar: I hope we can run daily (?) tests based on the latest RPMs, that would cover the -O2 case
09:21 ndevos we do have a (public facing) OpenStack instance at RH we can probably use for that
09:21 RaSTar ndevos: but isn't that catching the issue after merging the patch?
09:22 ndevos RaSTar: I very much doubt that we can catch issues with -O2 with our tests, it would mostly be some race conditions that may happen only rarely?
09:23 spalai joined #gluster-dev
09:23 RaSTar ndevos: oh no no, I am not interested to catch issues by wrong optimization due to O2. All this is just get discover all the glibc checks in regression.
09:24 RaSTar glibc is doing coverity like checks during run-time. If it finds an issue, it raises SIGABRT.
09:24 RaSTar Our community users see this.
09:24 RaSTar we don't see this in regression.
09:25 RaSTar because of the O0 flag and no FORTIFY
09:28 ndarshan joined #gluster-dev
09:32 ndevos RaSTar: right, so testing with RPMs would solve that?
09:34 RaSTar ndevos: yes
09:36 RaSTar ndevos: infact kshlm was suggesting that we build RPMs for every patch and then run regressions on them..
09:37 ndevos RaSTar: yeah, that would be nice too, I would like to see regression tests run with RPMs as well
09:37 ndevos actually, I could set that up in my Jenkins test env
09:37 RaSTar ndevos: I was not sure of how to get that done, so I added two of the RPM flags I know in build.sh
09:37 RaSTar ndevos: that would be awesome!
09:38 RaSTar ndevos: so these RPMs would be exactly like the ones we give out on download.gluster.org?
09:39 RaSTar ndevos: or more correctly like the ones that come from Fedora channels?
09:40 ndevos RaSTar: we could test the Fedora packages as well, and the ones from the CentOS SIG
09:40 ndevos RaSTar: but I prefer we start with the ones we build :)
09:40 RaSTar ndevos: no problem :) somewhere to start
09:41 kshlm ndevos, we already build rpms with the *devrpms jobs. Couldn't we just use them?
09:48 anrao joined #gluster-dev
09:52 ndevos kshlm: yes, we could, I've been experimenting with build-tarball -> build-rpms -> install-rpms-and-run-test in my own Jenkins installation
09:53 ndevos all of those are single jobs, and build-rpms is 3 concurrent ones, el5, el6 and el7, only when all succeed the install-rpms-and-run-test runs
09:54 ndevos jenkins is quite fun, but it takes a lot of time to try things out - which mostly will fail, and then you have to try again
10:06 spalai joined #gluster-dev
10:06 kd1 joined #gluster-dev
10:14 poornimag joined #gluster-dev
10:20 atinm joined #gluster-dev
10:21 pranithk xavih: I see you sent a massive patch :-)
10:22 pranithk xavih: I completed changes to ec_manager_access. Moving on to readdir[p]
10:35 anrao joined #gluster-dev
10:36 raghu joined #gluster-dev
10:39 xavih pranithk: It's not soo massive. There's a lot of dead code lines deleted :)
10:41 pranithk xavih: I will take a look as soon as I am done with my patch...
10:41 pranithk xavih: I wanted your inputs on http://review.gluster.com/11179
10:42 pranithk xavih: Bhaskar found one more issue where appending writes are leading to truncated file...
10:42 pranithk xavih: I mean if we write 100 lines, it ends up having 55 lines/39 lines etc
10:42 xavih pranithk: already reviewed the patch (I could review it for master, sorry)
10:42 pranithk xavih: I wasn't able to re-create that
10:43 pranithk xavih: Wow! nice
10:43 xavih pranithk: that's weird...
10:44 pranithk xavih: The only other problem that I don't know RCA is that renames are hanging on distributed disperse. We see that unlocks are happening on loc with null gfid in both inode and loc.gfid
10:44 pranithk xavih: I am yet to get RC that one as well
10:46 pranithk xavih: I will update if I find something...
10:46 xavih pranithk: thanks :)
10:47 atinm joined #gluster-dev
10:47 anekkunt joined #gluster-dev
10:59 pranithk xavih: one question though...
11:00 pranithk xavih: for access and readdir[p], winding on one of the good subvols is good enough IMO no need to wind on EC_MINIMUM_MIN
11:00 pranithk xavih: What do you feel?
11:12 xavih pranithk: readdir[p] will always wind on only one subvolume, access could do the same. However a previous xattrop will need to be send to all subvolumes to determine which ones are healthy
11:12 Gaurav__ joined #gluster-dev
11:14 pranithk xavih: yes, that is what I have done. Cool
11:14 xavih pranithk: :)
11:16 firemanxbr joined #gluster-dev
11:16 kd2 joined #gluster-dev
11:30 spalai joined #gluster-dev
11:45 hchiramm_ Manikandan++
11:45 glusterbot hchiramm_: Manikandan's karma is now 6
11:53 ndevos pranithk: did you check if a netbsd test was scheduled for the stack/list change? http://review.gluster.org/11095
11:54 * ndevos will do lunch now, back later!
11:55 Gaurav__ joined #gluster-dev
12:04 gem joined #gluster-dev
12:07 rjoseph joined #gluster-dev
12:07 anekkunt joined #gluster-dev
12:12 pranithk ndevos: no
12:16 itisravi_ joined #gluster-dev
12:23 ppai joined #gluster-dev
13:19 johnmark joined #gluster-dev
13:20 shyam joined #gluster-dev
14:17 pousley joined #gluster-dev
14:23 shubhendu joined #gluster-dev
14:23 wushudoin joined #gluster-dev
14:35 wushudoin joined #gluster-dev
15:01 nbalacha joined #gluster-dev
15:16 spalai joined #gluster-dev
15:34 hagarth joined #gluster-dev
15:34 krink joined #gluster-dev
15:44 firemanxbr_ joined #gluster-dev
15:57 shubhendu joined #gluster-dev
16:01 firemanxbr joined #gluster-dev
16:03 spalai joined #gluster-dev
16:19 spalai joined #gluster-dev
16:40 craigcabrey joined #gluster-dev
16:51 spalai joined #gluster-dev
17:08 josferna joined #gluster-dev
17:16 Gaurav__ joined #gluster-dev
17:40 anrao joined #gluster-dev
17:49 spalai joined #gluster-dev
18:02 lpabon joined #gluster-dev
18:16 hagarth joined #gluster-dev
18:20 shyam joined #gluster-dev
18:53 wushudoin| joined #gluster-dev
18:59 wushudoin| joined #gluster-dev
19:13 dlambrig1 left #gluster-dev
19:27 Susant_ left #gluster-dev
19:55 anrao joined #gluster-dev
20:21 dlambrig anyone know what this message from rfc.sh means: “error: The requested URL returned error: 406 Not Acceptable while accessing http://dlambrig@review.gluster.org/glusterfs.git/info/refs”
20:23 obnox dlambrig: you seem to have tried to post inappropriate content ... ;)
20:25 obnox s/th else: is the use of '!!' (double NOT) considered good style in gluster code?
20:25 dlambrig obnox: no idea what you are talking about.
20:26 obnox dlambrig: i just stumbled across an expression of the form !!foobar
20:26 obnox dlambrig: I had to think about it for a second
20:27 obnox dlambrig: there are several uses of that pattern around the code.
20:27 obnox dlambrig: for me it is a clever case of 'brevity over clarity' and I was wondering if that is considered good style
20:27 dlambrig obnox: hmm ok… (not related to my issue..)
20:28 obnox dlambrig: no, sorry. that was why I wrote 's/th else...'
20:29 dlambrig obnox: a double NOT seems unnecessary to me
20:31 obnox dlambrig: it turns !=0 into 1, basically
20:31 obnox e.g. !!(a pointer) will be 1 if the pointer is != NULL and 0 if the ptr == NULL
20:33 dlambrig can’t you just look the pointer unmodified?  if (ptr) { ptr not null} else {ptr is null}
20:33 obnox dlambrig: my question was rather: Oh, that is clever but really ugly. I wouldn't do that.
20:34 dlambrig fwiw .. seems ugly to me too.. but maybe it makes sense in how it is used in the code..
20:34 obnox it makes sense, but not everything that makes sense may be considered good style or so
20:36 dlambrig what file/line# is it used
20:36 obnox several: (says git grep '!!'):
20:36 obnox api/src/glfs-fops.c
20:36 obnox cli/src/cli-cmd-parser.c
20:36 obnox contrib/qemu/block.c
20:36 obnox contrib/fuse-lib/mount.c
20:36 obnox contrib/qemu/block/qed-check.c
20:37 obnox contrib/qemu/block/qed.c
20:37 obnox xlators/cluster/afr/src/afr-inode-write.c
20:37 obnox xlators/encryption/crypt/src/crypt.c
20:37 obnox xlators/meta/src/measure-file.c
20:37 obnox xlators/mgmt/glusterd/src/glusterd-volgen.c
20:37 obnox xlators/mount/fuse/src/fuse-bridge.c
20:37 obnox so there are a few
20:38 dlambrig if you remove the !!, a warning shows up “ warning: assignment makes integer from pointer without a cast”
20:38 dlambrig interesting
20:39 obnox :)
20:39 dlambrig I would rather have a macro “is_null(ptr)”
20:39 obnox dlambrig: but it is not related to pointers
20:40 obnox actually to any expression that you can use ! on
20:40 obnox s/related/restricted/
20:41 dlambrig is_nonzero( value ) for non pointers, is_null( ptr) for pointers
20:41 hagarth joined #gluster-dev
20:41 ndevos dlambrig: you have a "git remote -v" where origin is not a ssh URL?
20:42 dlambrig ndevos: its http, that must be the problem
20:43 ndevos dlambrig: yes, I think so, use " git remote set-url origin ssh://....."
20:43 dlambrig ndevos: cool
20:44 ndevos obnox: I'm not a fan of the !!something expression, but it is done in the linux kernel sources too, I would not bother changing it
20:45 obnox ndevos: ok.
20:45 * obnox shivers
20:45 * ndevos shrugs
20:46 obnox it is not as if something should be considered good style just because linux kernel does it ;)
20:46 ndevos haha, no, but it seems to be a common pattern?
20:47 ndevos feel free to change it into something else, if you like :)
20:47 obnox ndevos: maybe common. I have to admit I just conciously observed it for the first time.
20:47 obnox i was mainly curious.
20:48 ndevos obnox: I remember I has to think about it a little the first time I saw it too :)
20:48 obnox ndevos: that is exactly the point: brevity over clarity...
20:48 ndevos obnox: mabe #define to_bool(x)  (!!x)
20:52 * ndevos leaves for the day, ttyt!
20:53 obnox I somehow have the samba coding style guidlines rooted in my spinal cord. They say don't use (!ptr) , always use (ptr != NULL) ... are rooted
20:53 obnox ndevos: good n8
20:53 obnox s/are rooted//
20:54 badone_ joined #gluster-dev
20:54 ndevos oh, I think (ptr != NULL) is more used, if you want to change it, just send a patch :)
20:54 ndevos cya!
20:54 dlambrig +1 to macro over !!.
21:02 craigcabrey joined #gluster-dev
21:02 spalai joined #gluster-dev
21:15 anrao joined #gluster-dev
21:35 anrao joined #gluster-dev
21:46 krink joined #gluster-dev
22:02 surabhi joined #gluster-dev
22:10 craigcabrey joined #gluster-dev
23:02 shyam joined #gluster-dev
23:08 dlambrig left #gluster-dev
23:09 shyam joined #gluster-dev

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