#ceph IRC Log


IRC Log for 2010-07-23

Timestamps are in GMT/BST.

[0:13] * deksai (~chris@71-13-57-82.dhcp.bycy.mi.charter.com) has left #ceph
[2:26] * ghaskins_mobile (~ghaskins_@66-189-114-103.dhcp.oxfr.ma.charter.com) has joined #ceph
[2:38] * eternaleye (~quassel@184-76-53-210.war.clearwire-wmx.net) has joined #ceph
[2:47] * eternaleye_ (~quassel@184-76-53-210.war.clearwire-wmx.net) has joined #ceph
[2:48] <darkfader> "Image:Zabbix ceph templates.xml
[2:48] <darkfader> "
[2:48] * darkfader fails at attaching files
[2:49] * eternaleye (~quassel@184-76-53-210.war.clearwire-wmx.net) Quit (Ping timeout: 480 seconds)
[3:09] * eternaleye (~quassel@184-76-53-210.war.clearwire-wmx.net) has joined #ceph
[3:13] * eternaleye_ (~quassel@184-76-53-210.war.clearwire-wmx.net) Quit (Ping timeout: 480 seconds)
[4:54] * MarkN (~nathan@ Quit (Ping timeout: 480 seconds)
[5:07] * MarkN (~nathan@ has joined #ceph
[6:15] * Osso (osso@AMontsouris-755-1-7-189.w86-212.abo.wanadoo.fr) Quit (Quit: Osso)
[6:55] <f4m8_> alias OO /nick f4m8; wait 6000; /msg nickserv IDENTIFY boP#aeX4Koo}r9ge;
[6:55] * f4m8_ is now known as f4m8
[6:55] <f4m8> alias OO /msg nickserv IDENTIFY boP#aeX4Koo}r9ge;
[8:36] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has joined #ceph
[8:44] * eternaleye (~quassel@184-76-53-210.war.clearwire-wmx.net) Quit (Ping timeout: 480 seconds)
[8:53] * mtg (~mtg@vollkornmail.dbk-nb.de) has joined #ceph
[9:08] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) Quit (Quit: Yoric)
[9:20] * s15y (~s15y@sac91-2-88-163-166-69.fbx.proxad.net) Quit (Quit: WeeChat 0.2.6)
[10:08] * Yoric (~David@ has joined #ceph
[10:15] * f4m8 (~drehmomen@lug-owl.de) Quit (Quit: leaving)
[10:16] * f4m8 (~drehmomen@lug-owl.de) has joined #ceph
[10:17] * s15y (~s15y@sac91-2-88-163-166-69.fbx.proxad.net) has joined #ceph
[10:17] * allsystemsarego (~allsystem@ has joined #ceph
[12:28] * Osso (osso@AMontsouris-755-1-7-189.w86-212.abo.wanadoo.fr) has joined #ceph
[13:32] * Osso (osso@AMontsouris-755-1-7-189.w86-212.abo.wanadoo.fr) Quit (Quit: Osso)
[13:33] * f4m8 (~drehmomen@lug-owl.de) Quit (Quit: Changing server)
[13:33] * f4m8 (~drehmomen@lug-owl.de) has joined #ceph
[13:36] * f4m8 (~drehmomen@lug-owl.de) Quit ()
[13:36] * f4m8 (~drehmomen@lug-owl.de) has joined #ceph
[13:36] * f4m8 (~drehmomen@lug-owl.de) has left #ceph
[13:39] * f4m8 (~drehmomen@lug-owl.de) has joined #ceph
[13:40] * f4m8 (~drehmomen@lug-owl.de) Quit ()
[13:40] * f4m8 (~drehmomen@lug-owl.de) has joined #ceph
[13:56] * Yoric_ (~David@ has joined #ceph
[13:56] * Yoric (~David@ Quit (Read error: Connection reset by peer)
[13:56] * Yoric_ is now known as Yoric
[15:29] * mtg (~mtg@vollkornmail.dbk-nb.de) Quit (Quit: Verlassend)
[15:50] * deksai (~chris@71-13-57-82.dhcp.bycy.mi.charter.com) has joined #ceph
[18:14] * eternaleye (~quassel@184-76-53-210.war.clearwire-wmx.net) has joined #ceph
[18:39] * Yoric_ (~David@ has joined #ceph
[18:39] * Yoric (~David@ Quit (Read error: Connection reset by peer)
[18:39] * Yoric_ is now known as Yoric
[18:42] <yehudasa> wido: are you there?
[18:54] <wido> yes
[18:54] <wido> yehudasa: here now
[18:55] <yehudasa> I just updated #302
[18:55] <yehudasa> we do actually support if-modified-since
[18:56] <yehudasa> we probably just return a different error code
[18:57] <wido> but yes, Amazon does return a 304 in such case
[18:58] <wido> that's what the RFC specifies
[19:01] <yehudasa> hmm.. didn't find that 304 in their specifications
[19:01] <wido> did you check the RFC? http://www.w3.org/Protocols/rfc2616/rfc2616-sec14.html
[19:02] <wido> "If the variant has not been modified since a valid If- Modified-Since date, the server SHOULD return a 304 (Not Modified) response"
[19:05] <yehudasa> hmm.. I just looked at the code, we're actually supposed to return 304
[19:05] <yehudasa> so if it doesn't work it's probably some bug
[19:07] <wido> ok, didn't check it any further yet
[19:07] <wido> did you see my patch in #301?
[19:08] <yehudasa> wido: in rgw_rados.c, you can add prints, verify what's mtime and *mod_ptr are
[19:08] <yehudasa> might be some kind of a timezone shift issue
[19:08] <wido> for the last-modified? or the if-modified-since?
[19:08] <yehudasa> if-modified-since
[19:09] <yehudasa> in prepare_get_obj()
[19:09] <wido> yes, i thought so, but didn't have the time yet to check it
[19:09] <wido> so only made the issue and wanted to work on it later
[19:11] <wido> my patch won't be perfect, but i think it's a good start
[19:11] <wido> it works for me btw
[19:12] <yehudasa> haven't looked at it yet
[19:16] <yehudasa> wido: on what branch did you create this patch?
[19:20] <wido> rgw-206
[19:20] <yehudasa> yeah, I see that now
[19:20] <yehudasa> thanks
[19:28] <yehudasa> wido: two issues with the patches you sent
[19:28] <yehudasa> one is that we should dump the last modified if it wasn't initialized
[19:29] <yehudasa> and that can happen if we hit some error, I guess
[19:29] <yehudasa> and also, should use gmtime_r instead of gmtime
[19:29] <yehudasa> other than that it looks fine
[19:33] <wido> cool :)
[19:34] * Yoric (~David@ Quit (Quit: Yoric)
[19:37] <yehudasa> oh, I said "should", should be of course "shouldn't"
[19:42] <wido> btw, is the git web viewer working with you? i get a XML parsing error with firefox
[19:42] <sagewk> hmm, yep!
[19:43] <wido> ok, i'll check over here
[19:44] <wido> yehudasa: would it be an idea to make a seperate git repo for the gateway?
[19:44] <wido> it's not a part of Ceph is it? Like qemu-kvm with rbd
[19:44] <yehudasa> it's actually part of the ceph distribution
[19:44] <wido> ok
[19:45] <yehudasa> not sure whether we want to separate it, probably too much headache
[19:45] <wido> that could be right.
[19:46] <wido> about librados, we still have issues compiling without types.h
[19:46] <wido> haven't got the message, but it's something with snap_id
[19:46] <yehudasa> hmm.. I'll check that
[19:47] <yehudasa> I guess you have problem with librados.hpp?
[19:48] <wido> yes, we only try to initialize it and then close again
[19:48] <wido> simple main() function
[19:48] <yehudasa> I'll need the exact compilation error
[19:49] * deksai (~chris@71-13-57-82.dhcp.bycy.mi.charter.com) Quit (Ping timeout: 480 seconds)
[19:49] <wido> i get it, let me check if the PC where the code is, is still online
[19:52] <wido> yehudasa: http://www.pastebin.org/413537
[19:53] <wido> while it does when types.h is included from the source
[19:56] <yehudasa> ok, I'll fix that
[20:02] * deksai (~chris@71-13-57-82.dhcp.bycy.mi.charter.com) has joined #ceph
[20:19] <yehudasa> wido: pushed a fix to the librados problem, let me know if there are any other problems with that
[20:19] <yehudasa> also, I pushed your changes to the rgw-206 branch, but it still requires those two fixes
[20:38] <wido> yehudasa: tnx! i'll take a look at the rgw-206 branch for those two thing
[20:39] <wido> about the content length, could it be that the gateway actually doesn't output anything?
[20:39] <wido> so apache then return content-length: 0
[20:39] <wido> i saw the same today with Last-Modified, when that header was wrongly formatted, Apache overwrites it
[20:40] <yehudasa> wido: I'll reverify it again, but I don't think so
[20:40] <yehudasa> we're writing the header after there's available data
[20:40] <yehudasa> and if you replace the 206 with 200 it works
[20:40] <wido> ok, was just a thought.
[20:41] <wido> i've been searching around today on this, but nothing about Apache overwriting the header
[20:41] <yehudasa> yeah.. me neither
[20:45] <wido> i'll try lighttpd (if i get the rewrite rule working) to see what that does
[20:45] <wido> so we can rule out if it is Apache
[20:46] <yehudasa> originally we developed the rados gw on lighttpd, but it had some problems with sending 100-continue response
[20:46] <yehudasa> so we moved to apache
[20:46] <wido> how long ago was that?
[20:47] <yehudasa> hmm.. less than a year, probably ~8 months
[20:48] <wido> ok, thought it might be a old lighttpd bug, but 8 months is not so long ago
[20:49] <wido> but, FastCGI should work under any webserver ;)
[20:49] <yehudasa> yeah.. although there are two different implementations of it
[20:50] <yehudasa> the one we use is the apache fcgi module, not the fastcgi module
[20:51] <wido> that's true
[20:51] <wido> but i'll see if i can find a reason why 206 doesn't work and 200 does
[21:10] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has joined #ceph
[21:11] * Yoric (~David@dau94-10-88-189-211-192.fbx.proxad.net) has left #ceph
[21:24] * f4m8 (~drehmomen@lug-owl.de) Quit (reticulum.oftc.net coulomb.oftc.net)
[21:24] * f4m8 (~drehmomen@lug-owl.de) has joined #ceph
[21:25] * yehudasa (~yehudasa@ip-66-33-206-8.dreamhost.com) Quit (Read error: Connection reset by peer)
[21:35] * yehudasa (~yehudasa@ip-66-33-206-8.dreamhost.com) has joined #ceph
[21:49] <wido> i'm trying to change the replication level of my metdata pool to three (ceph osd pool set metadata size 3), the command goes fine, but nothing happends
[21:49] <wido> there is no data movement
[21:49] <wido> any ideas?
[21:50] <wido> btw, git seems down, get a connection reset by peer when i do a pull
[21:54] <gregaf> hmmm, I think metdata might have 3x replication by default
[21:54] <gregaf> not sure though
[21:54] <wido> no, it was on 2
[21:54] <wido> of that i'm sure
[21:55] <wido> i've just set "data" from 2 to 3, nothing happends either
[22:22] <yehudasa> wido: are we returning a different error code with the if-modified-since, or it's just being ignored?
[22:22] * darkfader (~floh@host-82-135-62-109.customer.m-online.net) Quit (Read error: Connection reset by peer)
[22:22] <wido> right now the reply is simply "200 OK"
[22:23] <wido> but there is more work to be done, when we return a 304, we should not return the Content-Length
[22:25] <yehudasa> wido: did you check why 200 OK is returned?
[22:28] <wido> not fully yet, but it seems that the gateway things there is data to return and then sends a 200 OK
[22:28] <wido> i don't see any structure for returning a 304
[22:31] <yehudasa> if (mod_ptr) {
[22:31] <yehudasa> if (mtime < *mod_ptr) {
[22:31] <yehudasa> err->num = "304";
[22:31] <yehudasa> err->code = "PreconditionFailed";
[22:31] <yehudasa> goto done_err;
[22:31] <yehudasa> }
[22:31] <yehudasa> }
[22:31] <yehudasa> that should have kicked in, at prepare_get_obj()
[22:32] <wido> yes, you are right. But are you sure it's PreconditionFailed? Since that is a 412
[22:33] <yehudasa> well.. that's just what'll be sent out in the xml error response
[22:35] <yehudasa> yeah, it should said NotModified
[22:35] <yehudasa> but it'll still return 304
[22:37] <yehudasa> the message was fixed and pushed
[22:39] <wido> ok, but still the 304 is not returned i think? btw, git is still down here, can't do a pull
[22:40] <yehudasa> it should return a 304.. if not, it's some other bug
[22:40] <wido> btw, it's also in rgw_fs.cc, same thing
[22:41] <yehudasa> yeah, you're right
[22:41] <yehudasa> need to really think whether we want to continue support the fs interface
[22:41] <yehudasa> it's probably broken anyway
[22:43] * yehudasa (~yehudasa@ip-66-33-206-8.dreamhost.com) Quit (Remote host closed the connection)
[23:18] * yehudasa (~yehudasa@ip-66-33-206-8.dreamhost.com) has joined #ceph
[23:18] <wido> yehudasa: i found the issue i think
[23:19] <yehudasa> what is it?
[23:19] <wido> in rgw_common.cc in the method parse_time the string was faulty
[23:19] <wido> uh, let me paste it
[23:19] <wido> http://www.pastebin.org/413787
[23:19] <wido> first, the string in was wrong
[23:20] <wido> and then the time had to be converted to a GMT time, since that's what HTTP uses for all its times
[23:20] <wido> now i'm getting a 304 Not Modified
[23:20] <yehudasa> yeah
[23:20] <yehudasa> the problem is that iirc, this is not a format that is compatible with amazon S3
[23:21] <wido> not in the XML, but it is for HTTP
[23:21] <wido> Amazon S3 also works with GMT times in their HTTP headers
[23:21] <wido> like the RFC specifies
[23:22] <yehudasa> alright
[23:22] <wido> can't create a patch right now btw, git pull still fails with Connection reset by peer
[23:22] <yehudasa> so we should probably have 2 different parse_time functions
[23:22] <yehudasa> one that's for the header, and one that's for the xml
[23:23] <wido> indeed, since they both use different formats
[23:24] <wido> but i don't see any XML times being parsed by parse_time? only the http errors
[23:24] <wido> uh, http headers
[23:27] <yehudasa> yeah.. I don't really remember where this format came from
[23:29] <yehudasa> oh, that's the iso 8601 date format
[23:30] <wido> so i think it's ok now, how the time is parsed
[23:31] <wido> i'm going afk
[23:32] <yehudasa> thanks!
[23:35] <wido> ok, ttyl! just uploaded a patch for #302
[23:35] <wido> << afk
[23:37] <yehudasa> hmm.. I just committed it from the patch on pastebin
[23:57] * yehudasa (~yehudasa@ip-66-33-206-8.dreamhost.com) Quit (Remote host closed the connection)

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