#ceph IRC Log


IRC Log for 2012-05-21

Timestamps are in GMT/BST.

[0:09] * The_Bishop (~bishop@cable-86-56-102-91.cust.telecolumbus.net) Quit (Ping timeout: 480 seconds)
[0:11] * s[X]_ (~sX]@eth589.qld.adsl.internode.on.net) has joined #ceph
[0:16] * BManojlovic (~steki@ Quit (Quit: Ja odoh a vi sta 'ocete...)
[0:27] * The_Bishop (~bishop@cable-86-56-102-91.cust.telecolumbus.net) has joined #ceph
[1:20] * aa (~aa@r190-135-2-207.dialup.adsl.anteldata.net.uy) Quit (Ping timeout: 480 seconds)
[1:29] * aa (~aa@r190-135-2-207.dialup.adsl.anteldata.net.uy) has joined #ceph
[2:09] * aa (~aa@r190-135-2-207.dialup.adsl.anteldata.net.uy) Quit (Ping timeout: 480 seconds)
[2:19] * yoshi (~yoshi@p3167-ipngn3601marunouchi.tokyo.ocn.ne.jp) has joined #ceph
[2:53] * aa (~aa@r190-135-2-207.dialup.adsl.anteldata.net.uy) has joined #ceph
[3:06] * aa (~aa@r190-135-2-207.dialup.adsl.anteldata.net.uy) Quit (Ping timeout: 480 seconds)
[3:24] * The_Bishop (~bishop@cable-86-56-102-91.cust.telecolumbus.net) Quit (Ping timeout: 480 seconds)
[3:35] * The_Bishop (~bishop@cable-86-56-102-91.cust.telecolumbus.net) has joined #ceph
[3:54] * aa (~aa@r190-135-2-207.dialup.adsl.anteldata.net.uy) has joined #ceph
[6:41] * f4m8_ is now known as f4m8
[7:37] * Theuni (~Theuni@ has joined #ceph
[7:51] * Theuni (~Theuni@ Quit (Ping timeout: 480 seconds)
[7:54] * Theuni (~Theuni@ has joined #ceph
[8:56] * Theuni (~Theuni@ Quit (Ping timeout: 480 seconds)
[9:07] * verwilst (~verwilst@d5152FEFB.static.telenet.be) has joined #ceph
[9:14] * s[X]_ (~sX]@eth589.qld.adsl.internode.on.net) Quit (Remote host closed the connection)
[9:27] * andreask (~andreas@chello062178013131.5.11.vie.surfer.at) has joined #ceph
[9:40] * BManojlovic (~steki@ has joined #ceph
[9:48] * todon (~todon@ has joined #ceph
[9:52] * ao (~ao@ has joined #ceph
[9:52] * pmjdebruijn (~pascal@overlord.pcode.nl) has joined #ceph
[9:52] <pmjdebruijn> hi guys
[9:52] <pmjdebruijn> http://ceph.newdream.net/download/ceph-0.47.tar.gz does not seem to exist yet
[9:53] <ao> Just wanted to post that. :)
[9:53] <ao> Should go into topic to avoid repeats.
[9:55] <filoo_absynth> oh, is .47 out?
[9:55] * andresambrois (~aa@r190-135-9-41.dialup.adsl.anteldata.net.uy) has joined #ceph
[9:56] * aa (~aa@r190-135-2-207.dialup.adsl.anteldata.net.uy) Quit (Read error: Connection reset by peer)
[9:56] * yoshi (~yoshi@p3167-ipngn3601marunouchi.tokyo.ocn.ne.jp) Quit (Read error: Connection reset by peer)
[9:56] <morpheus> filoo_absynth: http://ceph.com/releases/v0-47-released/
[9:56] * yoshi (~yoshi@p3167-ipngn3601marunouchi.tokyo.ocn.ne.jp) has joined #ceph
[11:03] * Theuni (~Theuni@ has joined #ceph
[11:12] * Theuni (~Theuni@ Quit (Quit: Leaving.)
[11:25] * MarkDude (~MT@c-71-198-138-155.hsd1.ca.comcast.net) Quit (Ping timeout: 480 seconds)
[11:48] * oliver (~oliver@p4FFFE7C1.dip.t-dialin.net) has joined #ceph
[11:52] * s[X]_ (~sX]@ppp59-167-157-96.static.internode.on.net) has joined #ceph
[12:32] * nhorman (~nhorman@99-127-245-201.lightspeed.rlghnc.sbcglobal.net) has joined #ceph
[13:15] * steki-BLAH (~steki@bojanka.net) has joined #ceph
[13:19] * BManojlovic (~steki@ Quit (Ping timeout: 480 seconds)
[13:50] * ThoughtCoder (ThoughtCod@60-240-78-43.static.tpgi.com.au) has joined #ceph
[13:51] <ThoughtCoder> Hey guys, I have created a new pool "backup" with ceph osd pool create backup'
[13:52] <ThoughtCoder> and I want to be able to mount this using cephfs - my understanding is that I need to run a 'ceph mds add_data_pool backup' is this still a valid command?
[13:53] <ThoughtCoder> My end goal is to be able to mount individual pools. I initially thought we could 'mount -t ceph /mnt/ceph/backup - but this hardlocks my server.
[13:53] <ThoughtCoder> Contrary to: http://ceph.com/wiki/FAQ#Can_I_mount_a_Ceph_file_system_at_an_arbitrary_depth.3F_.28i.e._mount_-t_ceph_11.111.111.111:.2Fsomedir.2F_.2Fmnt.2Fceph.29
[13:53] <ThoughtCoder> :(
[14:27] * verwilst (~verwilst@d5152FEFB.static.telenet.be) Quit (Quit: Ex-Chat)
[14:32] * ThoughtCoder (ThoughtCod@60-240-78-43.static.tpgi.com.au) Quit (Ping timeout: 480 seconds)
[14:59] * BManojlovic (~steki@ has joined #ceph
[15:02] * steki-BLAH (~steki@bojanka.net) Quit (Ping timeout: 480 seconds)
[15:17] * yoshi (~yoshi@p3167-ipngn3601marunouchi.tokyo.ocn.ne.jp) Quit (Remote host closed the connection)
[15:31] * xnox (~xnox@kursk.surgut.co.uk) has joined #ceph
[15:32] <xnox> On top of which filesystem is it currently recommended to run CEPH? btrfs?!
[15:32] <pmjdebruijn> last time we tried btrfs still had issue
[15:33] <pmjdebruijn> we are running it on ext4
[15:33] <pmjdebruijn> already we have a limited usecase
[15:33] <pmjdebruijn> although* (not already)
[15:33] <pmjdebruijn> xnox: stick around and see what others have to say
[15:33] <filoo_absynth> /sign
[15:33] * xnox will be idle here for a bit ;-)
[15:34] <filoo_absynth> tried btrfs, crashed badly. used ext4, crashed badly too, but it wasn't the FS's fault
[15:45] <ao> ext4 without journal just got a bug fixed in 3.2.18 that can hang osd. We are just now retrying with 3.2.18.
[15:48] <ao> btrfs also crashed here in all our tests.
[15:51] * f4m8 is now known as f4m8_
[16:26] <oliver> Hi... I will hope that there is no "mission-critical" reason, that 0.47 tarballs are not there yet ;) ?
[16:29] <filoo_absynth> just out of curiosity: who here is from germany, the netherlands, belgium or eastern france?
[16:30] * ao is from Germany.
[16:31] <pmjdebruijn> netherlands here
[16:34] <pmjdebruijn> filoo_absynth: why btw?
[16:41] <nhm> good morning #ceph
[16:42] <oliver> nhm: I have a wget-loop running for 0.47*.... Do you know, when it will be terminated?
[16:42] <oliver> nhm: good afternoon ;)=
[16:43] <nhm> oliver: when you stop it? ;)
[16:43] <oliver> Hehe
[16:44] <nhm> oliver: honestly no idea. I haven't been focusing on the releases much.
[16:44] <nhm> oliver: oh, supposedly it was released 8 hours ago!
[16:44] <nhm> http://ceph.com/releases/v0-47-released/
[16:45] <nhm> xnox: I'd give xfs a try.
[16:49] <oliver> True... but it's missing-in-action, uhm http://ceph.com/download/
[16:51] <nhm> doh!
[16:51] <nhm> let me see if I can do anything.
[16:56] * ao (~ao@ Quit (Quit: Leaving)
[16:57] <nhm> oliver: Looks like we'll need to wait until someone with access to that server gets in to fix it.
[16:58] <oliver> nhm: just trying to get it from git... describe get me a v0.47-55-gd8f4c6d. Is that OK?
[17:01] * adjohn (~adjohn@70-36-139-109.dsl.dynamic.sonic.net) has joined #ceph
[17:03] <nhm> oliver: probably? :)
[17:03] <nhm> oliver: honestly I didn't even realize we had released it until 15 mins ago.
[17:04] <oliver> nhm: ;) no prob... just doing dpkg-buildpackage... Will see what drops out... ( and what my test-environment tells me, harrr....)
[17:20] * andreask (~andreas@chello062178013131.5.11.vie.surfer.at) Quit (Ping timeout: 480 seconds)
[17:32] * bchrisman (~Adium@c-76-103-130-94.hsd1.ca.comcast.net) Quit (Quit: Leaving.)
[17:36] * Tv_ (~tv@aon.hq.newdream.net) has joined #ceph
[17:36] * BManojlovic (~steki@ Quit (Quit: Ja odoh a vi sta 'ocete...)
[17:38] <gregaf1> RupS: are your PGs still stuck?
[17:43] <oliver> nhm: hurray, it's there... Let the testing begin ;)
[17:45] * cattelan_away is now known as cattelan
[17:45] * MarkDude (~MT@c-71-198-138-155.hsd1.ca.comcast.net) has joined #ceph
[17:46] <nhm> oliver: excellent, good luck. :)
[17:47] <oliver> nhm: debian now updated, too, will by my way to do the installation. Thnx n bye for now ;)
[17:53] * Ryan_Lane (~Adium@208-117-193-99.static.idsno.net) has joined #ceph
[17:56] * andresambrois (~aa@r190-135-9-41.dialup.adsl.anteldata.net.uy) Quit (Remote host closed the connection)
[18:09] <sagewk1> elder: pushed a simple fix to testing, want to take a look?
[18:10] * oliver (~oliver@p4FFFE7C1.dip.t-dialin.net) has left #ceph
[18:15] * brambles (brambles@ Quit (Remote host closed the connection)
[18:16] * brambles (brambles@ has joined #ceph
[18:20] * BManojlovic (~steki@ has joined #ceph
[18:25] * andreask (~andreas@chello062178013131.5.11.vie.surfer.at) has joined #ceph
[18:36] * andreask (~andreas@chello062178013131.5.11.vie.surfer.at) Quit (Ping timeout: 480 seconds)
[18:38] <elder> sagewk1, the fix looks sound enough based on your explanation. I'll take a little closer look though.
[18:38] <elder> Ok, just looked closer:)
[18:38] <sagewk1> :) cool thanks
[18:38] * sagewk1 is now known as sagewk
[18:39] <elder> I'd prefer a (void) cast on the return to make it clear in this case you really mean to ignore the -ENOENT.
[18:39] <elder> In many cases I don't like such casts, but here I'd rather see it.
[18:41] * joao (~JL@aon.hq.newdream.net) has joined #ceph
[18:41] <sagewk> hmm, (void)foo(); or (void) foo(); ?
[18:42] * joao (~JL@aon.hq.newdream.net) Quit ()
[18:42] * joao (~JL@aon.hq.newdream.net) has joined #ceph
[18:43] <sagewk> elder: re-pushed.
[18:44] <elder> You I say (void) foo().
[18:44] <elder> But people differ.
[18:45] <elder> I adhere to the style I prefer when I can. I'm sure it annoys people.
[18:47] * The_Bishop (~bishop@cable-86-56-102-91.cust.telecolumbus.net) Quit (Quit: Wer zum Teufel ist dieser Peer? Wenn ich den erwische dann werde ich ihm mal die Verbindung resetten!)
[18:49] * nhorman (~nhorman@99-127-245-201.lightspeed.rlghnc.sbcglobal.net) Quit (Ping timeout: 480 seconds)
[18:51] * todon (~todon@ Quit (Quit: Ex-Chat)
[18:53] * rturk (~rturk@aon.hq.newdream.net) has joined #ceph
[18:53] * bchrisman (~Adium@ has joined #ceph
[18:54] * joshd (~joshd@aon.hq.newdream.net) has joined #ceph
[18:55] * nhorman (~nhorman@99-127-245-201.lightspeed.rlghnc.sbcglobal.net) has joined #ceph
[18:59] * Theuni (~Theuni@ has joined #ceph
[19:02] * chutzpah (~chutz@ has joined #ceph
[19:08] <yehudasa_> nhm: did you format the patched btrfs differently?
[19:08] <yehudasa_> nhm: also, 3.4 is out, we may want to switch to that
[19:10] <nhm> yehudasa_: I don't think so. Originally I was using btrfs dev but switched over to manually formatting and using fstab.
[19:10] <nhm> yehudasa_: why do you ask?
[19:12] <yehudasa_> nhm: performance degradation .. I think you need to format btrfs differently in order to take advantage of the new stuff
[19:12] <yehudasa_> e.g., mkfs.btrfs -l 64k -n 64k
[19:12] <nhm> yehudasa_: ah, ok
[19:13] <nhm> yehudasa_: Yeah, I was thinking about trying that anyway, didn't realize that's what you guys were talking about in the patched version.
[19:13] <nhm> Is that in 3.4 stock?
[19:14] <elder> Yes
[19:15] <elder> https://lkml.org/lkml/2012/5/20/126
[19:16] <nhm> cool
[19:24] * The_Bishop (~bishop@p4FCDF8BE.dip.t-dialin.net) has joined #ceph
[19:24] * Ryan_Lane (~Adium@208-117-193-99.static.idsno.net) Quit (Quit: Leaving.)
[19:34] * aa (~aa@r200-40-114-26.ae-static.anteldata.net.uy) has joined #ceph
[19:51] * lxo (~aoliva@lxo.user.oftc.net) Quit (Ping timeout: 480 seconds)
[19:52] * adjohn (~adjohn@70-36-139-109.dsl.dynamic.sonic.net) Quit (Quit: adjohn)
[19:54] * aa (~aa@r200-40-114-26.ae-static.anteldata.net.uy) Quit (Quit: Konversation terminated!)
[19:54] * aa (~aa@r200-40-114-26.ae-static.anteldata.net.uy) has joined #ceph
[19:57] * dmick (~dmick@aon.hq.newdream.net) has joined #ceph
[20:00] <gregaf1> nhm: how old are our long-lived test clusters right now, and what size are their mon's data directories? (alternatively: where do the mons live, and I'll go look)
[20:05] <sagewk> gregaf1: you can look at burnupi30
[20:06] * dmick (~dmick@aon.hq.newdream.net) has left #ceph
[20:08] * Ryan_Lane (~Adium@208-117-193-99.static.idsno.net) has joined #ceph
[20:09] <gregaf1> thanks
[20:12] <nhm> gregaf1: about 4-5 days old. Take a look at the "Congress Test Cluster Layouts" file in dreamhost google docs.
[20:13] * dmick (~dmick@aon.hq.newdream.net) has joined #ceph
[20:13] <nhm> gregaf1: Cluster 2 and 3 are the ones to look at.
[20:13] <gregaf1> burnupi30 looks to be almost a month old now and is at 765MB, that works for me :)
[20:13] * sage (~sage@cpe-76-94-40-34.socal.res.rr.com) Quit (Ping timeout: 480 seconds)
[20:17] <nhm> gregaf1: just glancing at one of the ones on the xfs cluster I've got, looks like it's about 480MB.
[20:18] <gregaf1> yeah, and the one I checked out it looks to be mostly in the log files
[20:18] <gregaf1> which can be rotated or whatever
[20:18] <nhm> yep, same here
[20:19] <nhm> log, log.debug, and log.info are all the exact same size (~122MB) and log.warn is slightly smaller.
[20:21] <rturk> I just got docs to build on a mac with macports. neat.
[20:23] * sage (~sage@cpe-76-94-40-34.socal.res.rr.com) has joined #ceph
[20:23] <gregaf1> oooh
[20:23] <gregaf1> push that somewhere accessible at some point :)
[20:23] <dmick> I guess I'm surprised the doc build uses that much that's nonportable
[20:24] <dmick> isn't it all, like, python and XML?
[20:24] <dmick> well and Doxygen, but still, all seemingly-portable userspace stuff
[20:24] * ThoughtCoder (~ThoughtCo@60-240-78-43.static.tpgi.com.au) has joined #ceph
[20:25] <rturk> and ditaa
[20:25] <rturk> # port install py-pip py-virtualenv libxml2 libxslt doxygen ditaa graphviz
[20:25] <rturk> # ln -s `which virtualenv-2.7` /usr/local/bin/virtualenv
[20:25] <rturk> # echo '#!/bin/sh' > /usr/local/bin/ditaa
[20:25] <rturk> # echo 'java -jar /opt/local/share/java/ditaa0_9.jar "$@"' >> /usr/local/bin/ditaa
[20:25] <rturk> that's how you make it work
[20:25] <rturk> not much to it, really
[20:26] <dmick> wow. Java just to do diagrams.
[20:26] <dmick> TIL
[20:26] <gregaf1> ah, simple enough
[20:27] <rturk> gregaf1: yep. most of that should probably be fixed in the ports themselves
[20:32] * ThoughtCoder (~ThoughtCo@60-240-78-43.static.tpgi.com.au) Quit (Ping timeout: 480 seconds)
[20:35] * Oliver (~oliver1@ip-109-90-142-176.unitymediagroup.de) has joined #ceph
[20:51] * s[X]_ (~sX]@ppp59-167-157-96.static.internode.on.net) Quit (Remote host closed the connection)
[20:53] * adjohn (~adjohn@ has joined #ceph
[21:12] * andreask (~andreas@chello062178013131.5.11.vie.surfer.at) has joined #ceph
[21:13] * MarkDude (~MT@c-71-198-138-155.hsd1.ca.comcast.net) Quit (Quit: Leaving)
[21:13] * MarkDude (~MT@c-71-198-138-155.hsd1.ca.comcast.net) has joined #ceph
[21:17] * lxo (~aoliva@lxo.user.oftc.net) has joined #ceph
[21:19] * adjohn (~adjohn@ Quit (Quit: adjohn)
[21:32] * sage (~sage@cpe-76-94-40-34.socal.res.rr.com) Quit (Ping timeout: 480 seconds)
[21:41] * sage (~sage@cpe-76-94-40-34.socal.res.rr.com) has joined #ceph
[21:47] * adjohn (~adjohn@ has joined #ceph
[21:50] <yehudasa_> elder: is there any xfs ioctl to defrag a directory metadata?
[21:51] <yehudasa_> swapext ioctl?
[22:22] <Tv_> yehudasa_: please see my email to the "ceph 0.47 libs3 install problem" thread
[22:24] <elder> yehudasa_, I don't think so. You can defragment a file but not a directory I don't think.
[22:24] <elder> And yes, swapext is how it's done.
[22:26] * joshd (~joshd@aon.hq.newdream.net) Quit (Ping timeout: 480 seconds)
[22:26] <nhm> elder: Any way to get xfs_db to give me min/max/stdev for directory/file fragmentation?
[22:26] <nhm> or dare I say a histogram? ;)
[22:28] <elder> Well that's a neat idea.
[22:28] <elder> But I think you need awk for that. Or perl or whatever your favorite scripting language. Python?
[22:29] <elder> Do you have something that's badly fragmented or something? It really shouldn't happen much.
[22:30] <elder> Where are these questions coming from?
[22:30] <nhm> elder: Yehuda and I are looking at the objects on the xfs ageing cluster.
[22:30] <dmick> yehudasa_ and nhm. :)
[22:30] * dmick ducks
[22:30] <elder> Ahhh.
[22:31] <nhm> elder: for the most part files are not fragmented (though thigns like the rgw logs have a lot of fragmentation). Yehuda found some directories with a lot of fragmentation though.
[22:31] <elder> Huh. What's a lot?
[22:31] <elder> Dozens? Hundreds? Thousands of extents?
[22:32] <nhm> elder: an example he gave me was 7 extents for 16 entries.
[22:32] <elder> That's weird.
[22:32] <nhm> overall I see:
[22:32] <nhm> xfs_db -r /dev/sdc1 -c 'frag -d'
[22:32] <nhm> actual 12740, ideal 8022, fragmentation factor 37.03%
[22:35] <dmick> anyone got a good filter rule for github review comments?
[22:35] <dmick> looking for something that really stands out
[22:35] <dmick> "Message-Id: <ceph/ceph/commit", maybe
[22:36] * joshd (~joshd@ has joined #ceph
[22:37] <dmick> eh. gmail wants to use list:(ceph.ceph.github.com) which might work too
[22:37] <gregaf1> dmick: you can probably just use "@reply.github.com"
[22:38] <gregaf1> I dump them to the same place as regular ceph commit messages though, so the list works just fine for me
[22:39] <dmick> that thing gmail suggested looks like it segregates ok
[22:40] <nhm> elder: Also, under what circumstances are the ideal number of directory extents 2? It seems to jump between 0 and 2 a lot when using verbose.
[22:40] * nhorman (~nhorman@99-127-245-201.lightspeed.rlghnc.sbcglobal.net) Quit (Quit: Leaving)
[22:41] <nhm> I rarely see an ideal of 1, and sometimes see 3 or 4.
[22:42] <elder> I haven't looked at this in great detail but as I recall, hitting the "ideal" is not that important, and even small amounts of fragmentation are not a big deal.
[22:42] <nhm> ok
[22:42] <elder> The only real problem is if you end up with like thousands of fragments.
[22:52] * cattelan (~cattelan@c-66-41-26-220.hsd1.mn.comcast.net) Quit (Quit: Terminated with extreme prejudice - dircproxy 1.2.0)
[22:57] * CristianDM (~CristianD@host217.190-230-240.telecom.net.ar) has joined #ceph
[22:58] * cattelan (~cattelan@c-66-41-26-220.hsd1.mn.comcast.net) has joined #ceph
[23:06] <CristianDM> If I put journaling into ramdisk, and server crash, is it possible recover it?
[23:07] <gregaf1> CristianDM: the server, or the journal?
[23:07] <gregaf1> the journal, no
[23:08] <CristianDM> journal
[23:08] <CristianDM> In Argentina the SSDs are too expensive :(
[23:08] <gregaf1> the server, if you are using btrfs, you can recover but the server won't be fully up-to-date so it'll need to be able to replicate current data from other servers (which means if you lose them all you will lose data)
[23:08] <gregaf1> if it were possible to recover data Ceph would use in-memory disks by default ;)
[23:08] <CristianDM> jeje :D
[23:09] <Tv_> if we only had non-volatile volatile memory...
[23:09] <gregaf1> nah, just non-volatile with volatile speed memory
[23:09] <gregaf1> it's coming???
[23:10] <CristianDM> Yes, but $$$
[23:10] <gregaf1> there's a thread on linux-fsdevel about the interface that should be used for it right now, actually
[23:11] <CristianDM> For now I will need to put journaling into normal HDD
[23:20] * Oliver (~oliver1@ip-109-90-142-176.unitymediagroup.de) has left #ceph
[23:23] * Theuni (~Theuni@ Quit (Ping timeout: 480 seconds)
[23:28] * aliguori (~anthony@cpe-70-123-145-39.austin.res.rr.com) Quit (Ping timeout: 480 seconds)
[23:31] * rturk1 (~rturk@aon.hq.newdream.net) has joined #ceph
[23:37] * rturk (~rturk@aon.hq.newdream.net) Quit (Ping timeout: 480 seconds)
[23:38] <nhm> gregaf1: have you looked at HP's memristor stuff?
[23:38] <gregaf1> nhm: uh, I know that HP et al have a thing called memristors that are dense, fast, and (will be Real Soon Now) cheap
[23:39] <gregaf1> and non-volatile
[23:39] * aliguori (~anthony@cpe-70-123-145-39.austin.res.rr.com) has joined #ceph
[23:41] <nhm> gregaf1: supposedly 2013.
[23:42] <nhm> http://www.bit-tech.net/news/hardware/2012/05/21/room-temp-reram/1
[23:51] * aliguori (~anthony@cpe-70-123-145-39.austin.res.rr.com) Quit (Read error: Connection reset by peer)
[23:52] * aliguori (~anthony@cpe-70-123-145-39.austin.res.rr.com) has joined #ceph

These logs were automatically created by CephLogBot on irc.oftc.net using the Java IRC LogBot.