#ceph IRC Log

Index

IRC Log for 2011-09-27

Timestamps are in GMT/BST.

[0:12] * bchrisman (~Adium@64.164.138.146) has joined #ceph
[0:24] * lxo (~aoliva@lxo.user.oftc.net) Quit (Remote host closed the connection)
[0:25] * lxo (~aoliva@lxo.user.oftc.net) has joined #ceph
[0:27] * adjohn is now known as Guest11678
[0:27] * adjohn (~adjohn@50.0.103.34) has joined #ceph
[0:31] * MarkN (~nathan@142.208.70.115.static.exetel.com.au) Quit (Quit: Leaving.)
[0:33] * Guest11678 (~adjohn@50.0.103.18) Quit (Ping timeout: 480 seconds)
[0:37] * bchrisman (~Adium@64.164.138.146) Quit (Quit: Leaving.)
[1:04] * jmlowe (~Adium@66.194.95.2) has joined #ceph
[1:17] * bchrisman (~Adium@64.164.138.146) has joined #ceph
[1:45] * bchrisman (~Adium@64.164.138.146) Quit (Quit: Leaving.)
[1:57] * cp (~cp@c-98-234-218-251.hsd1.ca.comcast.net) Quit (Quit: cp)
[2:04] * Tv|work (~Tv|work@aon.hq.newdream.net) Quit (Ping timeout: 480 seconds)
[2:05] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) Quit (Quit: Leaving)
[2:12] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) has joined #ceph
[2:23] * mtk (~mtk@ool-182c8e6c.dyn.optonline.net) Quit (Ping timeout: 480 seconds)
[2:26] * mtk (~mtk@ool-182c8e6c.dyn.optonline.net) has joined #ceph
[2:54] * gregorg_taf (~Greg@78.155.152.6) has joined #ceph
[2:54] * gregorg (~Greg@78.155.152.6) Quit (Read error: Connection reset by peer)
[3:29] * joshd (~joshd@aon.hq.newdream.net) Quit (Quit: Leaving.)
[3:43] * The_Bishop (~bishop@port-92-206-29-232.dynamic.qsc.de) has joined #ceph
[3:50] * adjohn (~adjohn@50.0.103.34) Quit (Quit: adjohn)
[4:56] * DanielFriesen (~dantman@S0106001731dfdb56.vs.shawcable.net) has joined #ceph
[4:56] * Dantman (~dantman@S0106001731dfdb56.vs.shawcable.net) Quit (Read error: Connection reset by peer)
[5:04] * yoshi (~yoshi@p9224-ipngn1601marunouchi.tokyo.ocn.ne.jp) has joined #ceph
[6:37] * adjohn (~adjohn@50-0-92-177.dsl.dynamic.sonic.net) has joined #ceph
[6:49] * bchrisman (~Adium@c-98-207-207-62.hsd1.ca.comcast.net) has joined #ceph
[6:59] * The_Bishop (~bishop@port-92-206-29-232.dynamic.qsc.de) Quit (Remote host closed the connection)
[8:25] * adjohn (~adjohn@50-0-92-177.dsl.dynamic.sonic.net) Quit (Quit: adjohn)
[9:26] * ghaskins_ (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) has joined #ceph
[9:27] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) Quit (Ping timeout: 480 seconds)
[9:30] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) has joined #ceph
[9:33] * ghaskins__ (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) has joined #ceph
[9:33] * adjohn (~adjohn@50-0-92-177.dsl.dynamic.sonic.net) has joined #ceph
[9:34] * ghaskins_ (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) Quit (Ping timeout: 480 seconds)
[9:39] * ghaskins (~ghaskins@66-189-113-47.dhcp.oxfr.ma.charter.com) Quit (Ping timeout: 480 seconds)
[10:34] * DanielFriesen (~dantman@S0106001731dfdb56.vs.shawcable.net) Quit (Remote host closed the connection)
[10:37] * Dantman (~dantman@S0106001731dfdb56.vs.shawcable.net) has joined #ceph
[10:44] * adjohn (~adjohn@50-0-92-177.dsl.dynamic.sonic.net) Quit (Quit: adjohn)
[10:52] * gohko (~gohko@natter.interq.or.jp) has joined #ceph
[11:23] * bchrisman (~Adium@c-98-207-207-62.hsd1.ca.comcast.net) Quit (Read error: Connection reset by peer)
[11:24] * bchrisman (~Adium@c-98-207-207-62.hsd1.ca.comcast.net) has joined #ceph
[13:00] * andret (~andre@pcandre.nine.ch) Quit (Ping timeout: 480 seconds)
[13:13] * andret (~andre@pcandre.nine.ch) has joined #ceph
[13:16] * yoshi (~yoshi@p9224-ipngn1601marunouchi.tokyo.ocn.ne.jp) Quit (Remote host closed the connection)
[13:44] * mtk (~mtk@ool-182c8e6c.dyn.optonline.net) Quit (Remote host closed the connection)
[13:49] * mtk (~mtk@ool-182c8e6c.dyn.optonline.net) has joined #ceph
[15:11] * andret (~andre@pcandre.nine.ch) Quit (Ping timeout: 480 seconds)
[15:16] * jmlowe (~Adium@66.194.95.2) Quit (Quit: Leaving.)
[16:57] * adjohn (~adjohn@50-0-92-177.dsl.dynamic.sonic.net) has joined #ceph
[17:14] * jmlowe (~Adium@66.194.95.2) has joined #ceph
[17:25] <df__> are there any tunables other than rsize for adjusting single threaded sequential read performance? setup: three OSDs, each with 10GbE. single kernel client (3.0.0) with 10GbE performing single threaded sequential reads only attains a peak of 1Gbit/sec.
[17:26] <df__> with a client on 1GbE, i'm getting about 500Mbit/sec in the same situation
[17:41] * bchrisman (~Adium@c-98-207-207-62.hsd1.ca.comcast.net) Quit (Quit: Leaving.)
[17:47] <gregaf1> df__: I think that's it unless you're on a very new kernel (containing commit 748725e8b9b7fa6ecea34798388dd7b9ffadff91, which I'm not sure has actually been sent to Linus)
[17:48] <df__> thats not in my current tree
[17:49] <gregaf1> it's probably just in ours
[17:49] <df__> curses kernel.org being down
[17:51] <gregaf1> if you're feeling brave you could try pulling down the newest Ceph stuff from git://github.com/NewDreamNetwork/ceph-client.git
[17:52] <gregaf1> I think that contains a bunch of readahead fixes even without adding a new tunable that make things much better
[17:52] <gregaf1> (50MB/s on a gigabit link is lame!)
[17:53] <df__> thanks, i'll give that a spin
[18:13] * aliguori (~anthony@cpe-70-123-132-139.austin.res.rr.com) Quit (Remote host closed the connection)
[18:18] * Tv|work (~Tv|work@aon.hq.newdream.net) has joined #ceph
[18:36] * jojy (~jojyvargh@108.60.121.114) has joined #ceph
[18:38] * lxo (~aoliva@lxo.user.oftc.net) Quit (Remote host closed the connection)
[18:39] * lxo (~aoliva@lxo.user.oftc.net) has joined #ceph
[18:40] * joshd (~joshd@aon.hq.newdream.net) has joined #ceph
[18:42] * aliguori (~anthony@32.97.110.59) has joined #ceph
[18:54] <sagewk> i think we should rename ceph-client-tools ceph-common and move more stuff in there
[18:54] <sagewk> notably, radosgw init script needs ceph-conf, which is currently in ceph
[18:54] <sagewk> and then make ceph, radosgw, etc. require ceph-common.
[19:02] * jmlowe (~Adium@66.194.95.2) Quit (Quit: Leaving.)
[19:09] <df__> gregaf1, i've done a quick test with for-linus and master, saw no change with the former, with master its doing ~2.8Gbit/sec single threaded read
[19:10] <gregaf1> sagewk will be glad it's better, even if it's not maxing out :/
[19:10] <df__> this is just using the default mount options
[19:10] <gregaf1> not really a surprise, though, I don't think we've ever tested on 10Gbe!
[19:10] <sagewk> df__: awesome. you should try adjusting the rasize mount option
[19:11] <gregaf1> you can probably do better by expanding rasize and rsize since you have so much more throughput
[19:17] * bchrisman (~Adium@108.60.121.114) has joined #ceph
[19:23] * adjohn (~adjohn@50-0-92-177.dsl.dynamic.sonic.net) Quit (Quit: adjohn)
[19:24] <df__> it'd be good if /proc/mounts enumerated the in use values for those options
[19:26] <sagewk> df__: hmm.. it spits them out only when they differ from the defaults.
[19:26] <sagewk> df__: i assumed that was the expected behavior, but it would be more useful it if showed you the current values unconditionally.
[19:29] <df__> there is good and bad about it. eg, you often end up with people copying /proc/mounts when setting NFS options in fstab, which has traditionally left people using very wrong rsize/wsize values -- for NFS it should be left to autonegotiate. on the other hand, its much simpler to know what the values are rather than having to look at how they are computed
[19:30] <Tv|work> there's also an argument for simple output there..
[19:31] <df__> personally, i'd rather it were all living in /sys nowerdays in a more general way. much like module options.
[19:41] <sagewk> we can easily add something to the debugfs dir that includes current options
[19:42] <sagewk> there's currently no ceph sysfs dir, so that'd be a bigger change
[20:03] * lxo (~aoliva@lxo.user.oftc.net) Quit (Quit: later)
[20:03] * lxo (~aoliva@lxo.user.oftc.net) has joined #ceph
[20:05] * jmlowe (~Adium@66.194.95.2) has joined #ceph
[20:31] * Coach007 (~coach007@p579B4815.dip.t-dialin.net) has joined #ceph
[20:48] * cp (~cp@c-98-234-218-251.hsd1.ca.comcast.net) has joined #ceph
[21:05] <Tv|work> sagewk: oh hey the start-stop-daemon thing we talked about i think yesterday.. "start-stop-daemon --start --exec /usr/bin/id -- -u -n"
[21:05] <sagewk> yeah, figured it out
[21:05] <Tv|work> sagewk: anything after -- is passed as arg
[21:05] <Tv|work> ok
[21:05] <Tv|work> i'm clearing out by scribbled notes ;)
[21:05] <sagewk> tnx
[21:05] <Tv|work> *my
[21:07] <sagewk> wanna talk cookbooks?
[21:07] <Tv|work> sagewk: i'm not sure i have enough to go yet, but if you have stuff then yes
[21:08] <sagewk> i have some stuff for the radosgw provisioning. but first pizza...
[21:08] <Tv|work> sagewk: radosgw is the easiest bit though
[21:09] <Tv|work> so i'm not really worried about that
[21:09] <Tv|work> it's intricate in apache2 versions etc, but it's integration with ceph is trivial
[21:09] <Tv|work> i'm more worried about osds and mons
[21:28] * jmlowe (~Adium@66.194.95.2) Quit (Quit: Leaving.)
[21:34] * ajm (adam@adam.gs) has left #ceph
[21:35] * ajm (adam@adam.gs) has joined #ceph
[21:51] * Coach007 (~coach007@p579B4815.dip.t-dialin.net) Quit (Ping timeout: 480 seconds)
[22:10] * DanielFriesen (~dantman@S0106001731dfdb56.vs.shawcable.net) has joined #ceph
[22:10] * Dantman (~dantman@S0106001731dfdb56.vs.shawcable.net) Quit (Read error: Connection reset by peer)
[22:13] <lxo> argh, root fs full because of ceph core files. any chance of making the dir for ceph core dumping configurable, so I could set it to a larger filesystem?
[22:14] <sjust> /proc/sys/kernel/core_pattern ?
[22:15] <lxo> can that change the directory?
[22:15] <gregaf1> that doesn't control location, just naming, right?
[22:15] <sjust> location as well
[22:15] <lxo> wow, nice, thanks
[22:15] <sjust> sure
[22:16] <lxo> though that will affect all core dumping, not just ceph's
[22:16] <sjust> yes
[22:16] * jmlowe (~Adium@66.194.95.2) has joined #ceph
[22:16] * jmlowe (~Adium@66.194.95.2) has left #ceph
[22:16] <sjust> though you can use the pattern to specify different directories for different binaries
[22:17] <sjust> something like /coredumps/%e/%p-core
[22:17] <sjust> where %e is the executable name and %p is the PID
[22:19] <lxo> and if the dir doesn't exist? no core file, or created on demand?
[22:19] <sjust> ah, that I don't know
[22:19] <lxo> kernel dumps core,then ;-)
[22:28] * cp (~cp@c-98-234-218-251.hsd1.ca.comcast.net) Quit (Quit: cp)
[23:20] * lx0 (~aoliva@lxo.user.oftc.net) has joined #ceph
[23:20] * lxo (~aoliva@lxo.user.oftc.net) Quit (Read error: Connection reset by peer)
[23:30] * aliguori (~anthony@32.97.110.59) Quit (Quit: Ex-Chat)
[23:31] * adjohn (~adjohn@ip-64-134-238-180.public.wayport.net) has joined #ceph
[23:32] <df__> hmm, ceph-client/master leaking memory?
[23:40] * mtk (~mtk@ool-182c8e6c.dyn.optonline.net) Quit (Remote host closed the connection)
[23:43] <Tv|work> sagewk: what exactly happens if i don't give --keyring= to cmon --mkfs?
[23:44] <Tv|work> sagewk: it seems like it "works", but does that mean it'll never ever accept any keys?
[23:44] <Tv|work> sagewk: does cmon *ever* read the keyrings mentioned in ceph.conf?
[23:52] * jojy_ (~jojyvargh@108.60.121.114) has joined #ceph
[23:52] * jojy (~jojyvargh@108.60.121.114) Quit (Read error: Connection reset by peer)
[23:52] * jojy_ is now known as jojy
[23:58] * aliguori (~anthony@cpe-70-123-132-139.austin.res.rr.com) has joined #ceph

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