| mattoliverau | lol | 00:00 |
|---|---|---|
| *** haomaiwang has joined #openstack-swift | 00:03 | |
| *** haomaiw__ has quit IRC | 00:03 | |
| *** Kennan has joined #openstack-swift | 00:12 | |
| Kennan | hi swift cores ? | 00:13 |
| Kennan | could you help one question ? | 00:13 |
| *** kota_ has joined #openstack-swift | 00:13 | |
| *** ChanServ sets mode: +v kota_ | 00:13 | |
| Kennan | http://lists.openstack.org/pipermail/openstack/2015-February/011502.html | 00:13 |
| Kennan | this seems not anwers why disable selinux work | 00:13 |
| *** haomaiwang has quit IRC | 00:13 | |
| Kennan | did not give explaniton | 00:14 |
| Kennan | could you give some explanation ? | 00:14 |
| *** dmorita has joined #openstack-swift | 00:29 | |
| mattoliverau | Kennan: I don't run swift on a centos/fedora/redhat, so don't know why selinux doesn't work, I assume there are no rules for the object-* to access /srv/node/sdc1/. Sounds like someone needs to write these, maybe via audit2allow or something. Maybe you need to wait until a Redhat swift core is in channel (probably tomorrow US time). | 00:33 |
| *** jrichli has joined #openstack-swift | 00:39 | |
| ho | Kennan: hello, I'm not a core but Remo answered your question in http://lists.openstack.org/pipermail/openstack/2015-February/011480.html . What kind of explanation do you want to have? (general behavior of selinux explanation) | 01:13 |
| Kennan | :ho I did not find it, why selinux enable can not work, what's the issue? | 01:14 |
| Kennan | many machines need to enable selinux for security | 01:15 |
| Kennan | I think disable selinux is not good to fix the issue | 01:15 |
| Kennan | :mattoliverau I am not sure who is redhat swift core | 01:15 |
| ho | Kennan: every processes on Linux which enables selinux need to handle it's permissions. it's not only swift processes but also other processes | 01:17 |
| Kennan | :ho so I am thinking how to solve that issue without disable selinux, as selinux is shared all system, it is not good to disable selinux just for swift | 01:18 |
| ho | Kennan: Ok, you would like to know how to put right permission to the processes? | 01:18 |
| Kennan | yes, I want to solve that issue without disable selinux | 01:19 |
| Kennan | ho do you know that? | 01:19 |
| ho | Kennan: which os do you use? rhel ? | 01:19 |
| Kennan | yes ho | 01:21 |
| ho | Kennan: wait a minute. I'm looking for an easy way to do it. | 01:22 |
| *** proteusguy has quit IRC | 01:26 | |
| *** thurloat is now known as thurloat_isgone | 01:29 | |
| ho | Kennan: install the package: https://repos.fedorapeople.org/repos/openstack/openstack-juno/epel-7/openstack-selinux-0.6.25-1.el7.noarch.rpm | 01:31 |
| Kennan | ho | 01:32 |
| Kennan | what's that package for ? | 01:33 |
| ho | Kennan: selinux configuration for openstack (including swift) | 01:33 |
| Kennan | ho I have installed that package | 01:33 |
| Kennan | rpm -qa have such package | 01:34 |
| ho | Kennan: you executed restorecon (I forgot correct name) in your env? | 01:34 |
| Kennan | yes, I executed on that direcotory, but still get permissioned denied issue | 01:35 |
| Kennan | ho so I did not know what's other solution except disable selinux? | 01:36 |
| ho | Kennan: is it possible paste your audit log which include AVC in the line? cat /var/log/audit/audit.log | grep "AVC" | 01:36 |
| Kennan | ho wait a minute | 01:37 |
| Kennan | ho | 01:39 |
| Kennan | herer is the log , let me try paste | 01:39 |
| *** proteusguy has joined #openstack-swift | 01:39 | |
| ho | Kennan: not directory paste here. you better to use paste site | 01:39 |
| Kennan | ho | 01:40 |
| Kennan | https://gist.github.com/HackToday/4eef80f198de62bd9fd7 | 01:40 |
| ho | Kennan: sorry... | 01:40 |
| Kennan | ho could you check the link? | 01:40 |
| Kennan | it should work | 01:40 |
| Kennan | ho have pasted one full log | 01:42 |
| Kennan | please check the same link | 01:42 |
| ho | Kennan: I can see it. I would like to have the result of cat /var/log/message | grep "setroubleshoot:" | 01:48 |
| Kennan | ho I did not find output for setroubleshoot: it is none match | 01:49 |
| Kennan | is it right keywold ? | 01:49 |
| ho | Kennan: i think so. no line? | 01:50 |
| Kennan | yes ho not any matched line | 01:50 |
| ho | Kennan: i see. I would like to know the process name for pid 6556 (Swift object auditor?) and usage of dm-2 device. | 01:54 |
| Kennan | ho the swift object auditor process is | 01:57 |
| Kennan | swift 6368 | 01:57 |
| Kennan | ho the dm-2 device | 01:57 |
| Kennan | is lrwxrwxrwx. 1 root root 7 Jun 14 20:51 swifttest -> ../dm-2 | 01:57 |
| Kennan | I have a disk | 01:57 |
| Kennan | and partition it used for swift storage device | 01:58 |
| Kennan | ho this is the example http://docs.openstack.org/kilo/install-guide/install/apt/content/swift-install-storage-node.html | 01:58 |
| ho | Kennan: I would like to know the result of ls -Z / | 01:58 |
| ho | Kennan: I have no rhel env so the option might be wrong ... | 01:59 |
| Kennan | ho https://gist.github.com/HackToday/4eef80f198de62bd9fd7 | 01:59 |
| Kennan | have added the link | 01:59 |
| Kennan | ho the issue really maybe selinux, my steps as following: 1, have an extra disk ,and format it, 2. config it used for swift mkfs.xfs /dev/sdb1 , mkdir -p /srv/node/sdb1, mount /srv/node/sdb1 | 02:01 |
| Kennan | ho, the issue is that when swift upload command | 02:01 |
| Kennan | it have complainted for | 02:01 |
| Kennan | Permission denied: '/srv/node/sdb1/accounts' | 02:01 |
| Kennan | it seems the http://docs.openstack.org/kilo/install-guide/install/apt/content/swift-install-storage-node.html | 02:02 |
| ho | Kennan: sorry I would like to know the result of ls -Z (for dm-2 device) | 02:02 |
| Kennan | not talked about selinux | 02:02 |
| Kennan | ho brw-rw----. root disk system_u:object_r:fixed_disk_device_t:s0 /dev/dm-2 | 02:02 |
| ho | Kennan: the seculity context on the device is the problem which you met. | 02:03 |
| Kennan | ho yes, I am thinking that, but i did not what's wrong with that setting ? I just followed community guide to do that :( | 02:03 |
| ho | Kennan: the link which you pasted is for ubuntu. | 02:07 |
| Kennan | yes ho, the rheal nearly have same, just replace apt with yum | 02:08 |
| Kennan | let me paste your rhel | 02:08 |
| Kennan | ho this is yum | 02:08 |
| Kennan | http://docs.openstack.org/kilo/install-guide/install/yum/content/swift-install-storage-node.html | 02:09 |
| ho | Kennan: i see. | 02:15 |
| ho | Kennan: maybe I mis-understood. ls -Z /srv/node | 02:19 |
| Kennan | ho OK I also not know what's wrong with that. Community guide not talke abut selinux, and its setting trouble for permission denied solution :0 | 02:21 |
| ho | Kennan: can i have an output of "ls -Z /srv/node" | 02:25 |
| Kennan | ho: | 02:27 |
| Kennan | drwxr-xr-x. swift swift system_u:object_r:unlabeled_t:s0 sdb1 | 02:27 |
| ho | Kennan: a redhat core for specialist of this is zaitcev (Pete Zaitev) | 02:28 |
| ho | Kennan: execute " restorecon -R /srv/node" and then "ls -Z /srv/node" | 02:29 |
| Kennan | ho: check this now | 02:30 |
| Kennan | drwxr-xr-x. swift swift system_u:object_r:swift_data_t:s0 sdb1 | 02:30 |
| ho | Kennan: Great! execute your failed operation again. | 02:31 |
| Kennan | ho: yeap it works now | 02:32 |
| ho | Kennan: Cool! everything works for you? | 02:33 |
| Kennan | ho so the reason should be swift_data_t ? is it swift selinux setting ? | 02:33 |
| ho | Kennan: rhel's selinux setting for swift. you installed rhel selinux pkg, this pkg configures selinux context with swift_xxx. | 02:36 |
| Kennan | ho OK Thanks. Glad to know that. :) | 02:37 |
| ho | Kennan: you are welcome. have fun with swift :-) | 02:38 |
| *** bkopilov has quit IRC | 02:40 | |
| mattoliverau | ho: nice work :) | 02:47 |
| kota_ | ho: nice | 02:49 |
| ho | mattoliverau, kota_: thanks! Kennan mentioned we have some problem in the doc. | 02:57 |
| *** silor has joined #openstack-swift | 03:13 | |
| mattoliverau | Yeah, we may need to mention you need to run restorecon if you are running a redhat variant and selinux | 03:21 |
| *** ronenkat has quit IRC | 03:25 | |
| *** logan2 has quit IRC | 03:35 | |
| *** jrichli has quit IRC | 03:38 | |
| *** bkopilov has joined #openstack-swift | 03:55 | |
| *** bkopilov has quit IRC | 04:06 | |
| *** bkopilov has joined #openstack-swift | 04:18 | |
| *** asettle is now known as asettle-afk | 04:25 | |
| *** ppai has joined #openstack-swift | 04:37 | |
| ho | acoles: I summarized the keystonemiddleware issue. http://paste.openstack.org/show/292901/ | 04:59 |
| ho | acoles: I used a backward comati issue for L21-24. | 05:00 |
| ho | acoles: I think I should write spec for "oslo config suports in swift". what do you think? | 05:00 |
| *** kbee has joined #openstack-swift | 05:02 | |
| *** proteusguy has quit IRC | 05:07 | |
| *** asettle-afk is now known as asettle | 05:10 | |
| *** proteusguy has joined #openstack-swift | 05:23 | |
| Kennan | ho there? | 05:25 |
| *** SkyRocknRoll has joined #openstack-swift | 05:28 | |
| *** ronenkat has joined #openstack-swift | 05:31 | |
| ho | Kennan: yes but i'm in a meeting now. | 05:52 |
| Kennan | ho OK I setup another environment, this environment is different from the one we discussed this morning | 05:56 |
| Kennan | I used loop device | 05:56 |
| Kennan | but I found even restorecon -R can not change that permission | 05:56 |
| *** Kennan2 has joined #openstack-swift | 06:03 | |
| *** Kennan has quit IRC | 06:04 | |
| ho | Kennan: you mounted the loop back device from /srv/node/xxx? | 06:04 |
| Kennan2 | ho | 06:14 |
| *** Kennan2 is now known as Kennan | 06:14 | |
| Kennan | I have setup like this | 06:14 |
| Kennan | I used trunate to create file like swift.img is 1G | 06:15 |
| Kennan | and /etc/fstab have mount like this | 06:15 |
| Kennan | for this /etc/swift/data/drives/images/swift.img /etc/swift/data/drives/sdb1 xfs loop,noatime,nodiratime,nobarrier,logbufs=8 0 0 | 06:15 |
| Kennan | then configure in swift about devices like this | 06:16 |
| Kennan | devices = /etc/swift/data/drives | 06:16 |
| Kennan | I tried for restrorecon on /etc/swift/data/drives | 06:17 |
| Kennan | but it can not work | 06:17 |
| Kennan | still have permission denied | 06:17 |
| Kennan | ls -lZ /etc/swift/data/drives/ | 06:17 |
| Kennan | drwxr-xr-x. swift swift system_u:object_r:etc_t:s0 sdb1 | 06:18 |
| ho | Kennan: it seems target of selinux pkg is /srv/node other directory is not supported. | 06:18 |
| Kennan | ho: is it ? I did not know that, I thought /srv/node is just one example | 06:19 |
| ho | Kennan: i'm not exactly sure but i found the https://github.com/redhat-openstack/openstack-selinux/blob/el6/openstack-selinux-swift.fc#L41 | 06:42 |
| ho | Kennan: but this is for rhel6 | 06:42 |
| Kennan | ok ho: let me try srv/node for this loop device case | 06:43 |
| *** ian_brown has quit IRC | 06:43 | |
| Kennan | ho: from your link, it seems redhat 6 have that, but I did not know what's setting for redhat 7? | 06:49 |
| Kennan | is it same as that ? i install on redhat 7.1 | 06:50 |
| ho | Kennan: i coudn't find it. they changed it b/w tag 0.15 and 0.5.1. | 06:51 |
| *** bkopilov has quit IRC | 06:56 | |
| ho | Kennan: I don't know the rhel7 config but i think it might work : in fstab /etc/swift/data/drives/images/swift.img /etc/swift/data/drives/sdb1 xfs noatime,nodiratime,nobarrier,logbufs=8,context=system_u:object_r:swift_data_t:s0 0 0 | 06:56 |
| ho | Kennan: change mount option. | 06:56 |
| Kennan | ho OK I would try that, BTW, i tried /srv/node just now, it worked with restroecon ways | 06:57 |
| Kennan | so not sure if /srv/node is hardcode or something in redhat | 06:58 |
| Kennan | for your suggested new mount option, I will try it | 06:58 |
| ho | Kennan: thanks! i think it's hardcoded. we will ask this to zaitcev :-) | 07:00 |
| cschwede | Good Morning! | 07:14 |
| cschwede | ho: Kennan: I don’t think it is hardcoded; it’s a default that you can set in [account|container|object]-server.conf. For example: https://github.com/openstack/swift/blob/master/etc/account-server.conf-sample#L8 | 07:14 |
| Kennan | cschwede We configured that, the issue is that selinux has some trick points | 07:16 |
| cschwede | Kennan: ah, you mean the value is hardcoded in the selinux package? | 07:16 |
| Kennan | yes, I just guess, I am trying ho: suggestion, I tried to change mount option | 07:17 |
| Kennan | as I have verified that other directorys like /etc/swift/data/drives can not works | 07:17 |
| Kennan | but /srv/node can work | 07:18 |
| cschwede | Kennan: looking at the .src.rpm I think it works for everything under /srv | 07:20 |
| cschwede | and /var/run/swift too | 07:20 |
| mattoliverau | cschwede: morning | 07:22 |
| cschwede | Morning Matthew! | 07:24 |
| Kennan | ho: I tried your suggestion it works | 07:25 |
| Kennan | so it seems mount need to specify that selinux context | 07:26 |
| ho | Kennan: Great! yeah if the /srv/node is hardcoded, adding security context of mount will be a solution. :-) | 07:29 |
| Kennan | thanks ho: your suggestion really helpful | 07:29 |
| cschwede | ho: looks like a nice tip, maybe you want to add this to the docs? | 07:30 |
| mattoliverau | +1 | 07:33 |
| ho | cschwede: morning! users can add own security context on their directory so I'm not sure which is a preferable way | 07:34 |
| *** chlong has quit IRC | 07:34 | |
| ho | Kennan: you are welcome! | 07:35 |
| *** acoles_away is now known as acoles | 07:37 | |
| acoles | good morning | 07:38 |
| mattoliverau | acoles: morning sir | 07:39 |
| mattoliverau | K, I need to go cook some dinner, probably be back later :) | 07:40 |
| *** joeljwright has joined #openstack-swift | 07:54 | |
| ho | acoles: morning! I wrote comments three hours before. | 07:59 |
| *** geaaru has joined #openstack-swift | 08:00 | |
| *** bkopilov has joined #openstack-swift | 08:06 | |
| *** joeljwright has quit IRC | 08:11 | |
| *** kbee has quit IRC | 08:13 | |
| acoles | ho: hi! i read your pastebin. i think what you are suggesting is that we could *force* "project = swift" to be fixed (user cannot change) by using your patch in keystoneauth. But don't we still have other config values for authtoken that may conflict between the paste ini values and a (future) oslo config in swift? | 08:25 |
| acoles | ho: Also, if "project=swift" is set in CONF in keystoneauth, does that happen *before* authtoken attempts to read CONF or is it too late? IDK. | 08:26 |
| *** slo_ has joined #openstack-swift | 08:28 | |
| *** joeljwright has joined #openstack-swift | 08:44 | |
| *** joeljwright1 has joined #openstack-swift | 08:51 | |
| ho | acoles: (1) I think my patch can change proxy-server.conf-sample but it's difficult to change existed proxy-server.conf. | 08:51 |
| ho | acoles: I think we don't have other conflicts. There is a difference of deployment (paste init or oslo config) but we use same info (which exposes to users). We just decide a rule which info swift will use. Current concern is 'project' is an internal info in OpenStack but only swift exposes it so we have a conflict. | 08:51 |
| ho | acoles: Keystoneauth is configured after authtoken so that means keystoneauth initializes first I think. | 08:52 |
| *** joeljwright has quit IRC | 08:52 | |
| acoles | ho: yes i think the middlewares are loaded in reverse order so keystoneauth would be init'd before auth_token. | 09:03 |
| acoles | ho: so is your concern is that "project = swift" should NEVER be something that users can change? or am i confused? :) | 09:05 |
| *** silor has quit IRC | 09:06 | |
| *** krykowski has joined #openstack-swift | 09:07 | |
| acoles | ho: or is that you think 'project' should ONLY be set using oslo.config and not set in paste deploy ini? | 09:11 |
| acoles | ho : hmmm, I can't see 'project' defined in the authtoken middleware _OPTs, maybe it is defined elsewhere. | 09:12 |
| *** tellesnobrega has quit IRC | 09:24 | |
| *** proteusguy has quit IRC | 09:25 | |
| *** kbee has joined #openstack-swift | 09:25 | |
| *** foexle has joined #openstack-swift | 09:34 | |
| ho | acoles: I like 'project' should ONLY be set using oslo.config and not set in paste deploy ini, if swift will support oslo config in near future. | 09:37 |
| *** haomaiwang has joined #openstack-swift | 09:38 | |
| ho | acoles: fyi: https://github.com/openstack/oslo.config/blob/master/oslo_config/cfg.py#L1830 | 09:40 |
| *** proteusguy has joined #openstack-swift | 09:40 | |
| *** jordanP has joined #openstack-swift | 09:43 | |
| acoles | ho: thanks for that link! ^^ | 09:50 |
| *** Kennan2 has joined #openstack-swift | 10:03 | |
| *** Kennan has quit IRC | 10:05 | |
| *** ianbrown has joined #openstack-swift | 10:07 | |
| *** ho has quit IRC | 10:07 | |
| acoles | ho: updated my review, but I am still not sure!) | 10:08 |
| *** aix has joined #openstack-swift | 10:10 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/191730 | 10:13 |
| *** marzif_ has joined #openstack-swift | 10:18 | |
| *** ianbrown has quit IRC | 10:33 | |
| *** marzif_ has quit IRC | 10:41 | |
| *** marzif_ has joined #openstack-swift | 10:41 | |
| *** dmorita has quit IRC | 10:45 | |
| *** kbee has quit IRC | 10:46 | |
| *** krykowski has quit IRC | 10:48 | |
| *** tellesnobrega has joined #openstack-swift | 10:57 | |
| *** krykowski has joined #openstack-swift | 11:01 | |
| acoles | ho: re. writing a spec for oslo.config in swift, it might be helpful to understand what the roadmap would be to move that way, in particular if we can have a hybrid for transition, but note concerns documented here: https://etherpad.openstack.org/p/swift_gap_scratchpad | 11:07 |
| acoles | ho: NB that etherpad has been inactive for 10 months. | 11:07 |
| acoles | ho: also this was torgomatic patch for keystonemiddleware https://review.openstack.org/#/c/152283/ | 11:14 |
| *** marzif_ has quit IRC | 11:24 | |
| *** marzif_ has joined #openstack-swift | 11:27 | |
| *** aix has quit IRC | 11:28 | |
| *** kota_ has quit IRC | 11:33 | |
| *** shooding has joined #openstack-swift | 11:48 | |
| shooding | Hello guys. I've a question about overload in the design of swift ring builder. | 11:49 |
| shooding | In order to respect weight of devices, a partition is mapped to only 2 device and what happens when overload is enabled? | 11:52 |
| *** kei_yama has quit IRC | 11:54 | |
| *** silor has joined #openstack-swift | 12:01 | |
| *** km has quit IRC | 12:03 | |
| *** ppai has quit IRC | 12:10 | |
| *** aix has joined #openstack-swift | 12:15 | |
| *** vinsh has quit IRC | 12:18 | |
| *** vinsh has joined #openstack-swift | 12:18 | |
| *** chlong has joined #openstack-swift | 12:21 | |
| *** vinsh has quit IRC | 12:22 | |
| *** ppai has joined #openstack-swift | 12:23 | |
| *** thurloat_isgone is now known as thurloat | 12:29 | |
| *** ppai has quit IRC | 12:31 | |
| *** marzif_ has quit IRC | 12:33 | |
| *** ptb has joined #openstack-swift | 12:35 | |
| *** ptb has quit IRC | 12:37 | |
| *** shooding has quit IRC | 12:37 | |
| *** RackerShagz has joined #openstack-swift | 12:38 | |
| *** RackerShagz has quit IRC | 12:42 | |
| *** ronenkat has quit IRC | 12:46 | |
| *** thurloat is now known as thurloat_isgone | 12:52 | |
| *** petertr7_away is now known as petertr7 | 12:53 | |
| *** bill_az has joined #openstack-swift | 13:03 | |
| *** mwheckmann has joined #openstack-swift | 13:04 | |
| *** petertr7 is now known as petertr7_away | 13:14 | |
| openstackgerrit | Stuart McLaren proposed openstack/python-swiftclient: Add minimal working service token support. https://review.openstack.org/182640 | 13:15 |
| *** mwheckmann has quit IRC | 13:20 | |
| *** vinsh has joined #openstack-swift | 13:24 | |
| *** vinsh has quit IRC | 13:25 | |
| *** vinsh has joined #openstack-swift | 13:25 | |
| *** mwheckmann has joined #openstack-swift | 13:26 | |
| openstackgerrit | Stuart McLaren proposed openstack/python-swiftclient: Add minimal working service token support. https://review.openstack.org/182640 | 13:27 |
| *** vinsh has quit IRC | 13:29 | |
| *** bi_fa_fu has joined #openstack-swift | 13:31 | |
| *** acampbell has joined #openstack-swift | 13:31 | |
| *** petertr7_away is now known as petertr7 | 13:31 | |
| *** SkyRocknRoll has quit IRC | 13:43 | |
| *** krykowski has quit IRC | 13:50 | |
| *** ronenkat has joined #openstack-swift | 13:53 | |
| *** jrichli has joined #openstack-swift | 14:02 | |
| *** blmartin has joined #openstack-swift | 14:04 | |
| *** kbee has joined #openstack-swift | 14:15 | |
| *** openstackgerrit has quit IRC | 14:24 | |
| *** openstackgerrit has joined #openstack-swift | 14:24 | |
| *** mwheckmann has quit IRC | 14:34 | |
| *** petertr7 is now known as petertr7_away | 14:38 | |
| *** blmartin_ has joined #openstack-swift | 14:40 | |
| *** mwheckmann has joined #openstack-swift | 14:41 | |
| *** amoturi has joined #openstack-swift | 14:41 | |
| *** blmartin has quit IRC | 14:44 | |
| *** blmartin_ has quit IRC | 14:44 | |
| *** blmartin has joined #openstack-swift | 14:45 | |
| *** marcusvrn has joined #openstack-swift | 14:45 | |
| *** minwoob has joined #openstack-swift | 14:50 | |
| *** CR7 has joined #openstack-swift | 14:51 | |
| *** breitz has joined #openstack-swift | 14:52 | |
| *** diazjf has joined #openstack-swift | 14:56 | |
| *** haomaiw__ has joined #openstack-swift | 15:03 | |
| *** haomaiwang has quit IRC | 15:04 | |
| *** petertr7_away is now known as petertr7 | 15:06 | |
| *** mwheckmann has quit IRC | 15:09 | |
| *** haomaiwa_ has joined #openstack-swift | 15:11 | |
| *** haomaiw__ has quit IRC | 15:13 | |
| *** acoles is now known as acoles_away | 15:13 | |
| zigo | Is there anyone besides Tushar able to fix pyeclib? | 15:18 |
| zigo | The issue has been pending for days already ... | 15:18 |
| zigo | I'm talking about https://bitbucket.org/kmgreen2/pyeclib/issue/64/wrong-version | 15:19 |
| *** CR7 has quit IRC | 15:22 | |
| *** kbee has quit IRC | 15:24 | |
| cschwede | peluse: could you ping Tushar or Kevin to help fix this packaging issue? It seems both are not in the channel atm ^^ | 15:27 |
| cschwede | zigo: Kevin might be able to help fix this as well; not sure if the email is still up2date, but have a look at his address in setup.py | 15:28 |
| zigo | cschwede: Thanks Christian! :) | 15:29 |
| zigo | I'll try to be more patient. | 15:29 |
| cschwede | zigo: you’re welcome Thomas :) | 15:30 |
| *** foexle has quit IRC | 15:31 | |
| notmyname | good morning | 15:32 |
| *** jordanP has quit IRC | 15:37 | |
| *** blmartin_ has joined #openstack-swift | 15:47 | |
| *** blmartin has quit IRC | 15:47 | |
| *** blmartin__ has joined #openstack-swift | 15:49 | |
| *** blmartin_ has quit IRC | 15:49 | |
| *** bi_fa_fu has quit IRC | 15:50 | |
| *** petertr7 is now known as petertr7_away | 15:56 | |
| *** amoturi has quit IRC | 16:01 | |
| *** barker has joined #openstack-swift | 16:15 | |
| MooingLemur | gmornin | 16:26 |
| *** diazjf1 has joined #openstack-swift | 16:29 | |
| *** diazjf has quit IRC | 16:29 | |
| *** gyee_ has joined #openstack-swift | 16:30 | |
| *** diazjf1 has quit IRC | 16:31 | |
| *** barker has quit IRC | 16:31 | |
| *** blmartin_ has joined #openstack-swift | 16:35 | |
| *** blmartin__ has quit IRC | 16:35 | |
| *** janonymous_ has joined #openstack-swift | 16:38 | |
| *** blmartin_ has quit IRC | 16:38 | |
| openstackgerrit | janonymous proposed openstack/swift: Replace it.next() with next(it) for py3 compat https://review.openstack.org/191881 | 16:44 |
| *** thurloat_isgone is now known as thurloat | 16:47 | |
| *** blmartin has joined #openstack-swift | 16:47 | |
| *** petertr7_away is now known as petertr7 | 16:47 | |
| *** fthiagogv has joined #openstack-swift | 16:56 | |
| *** fthiagogv_ has joined #openstack-swift | 16:56 | |
| *** amoturi has joined #openstack-swift | 16:56 | |
| *** petertr7 is now known as petertr7_away | 16:57 | |
| *** mmcardle1 has quit IRC | 16:57 | |
| *** haomaiw__ has joined #openstack-swift | 17:03 | |
| *** haomaiwa_ has quit IRC | 17:04 | |
| *** joeljwright1 has quit IRC | 17:06 | |
| *** petertr7_away is now known as petertr7 | 17:08 | |
| *** lcurtis has joined #openstack-swift | 17:09 | |
| *** MooingLemur has quit IRC | 17:13 | |
| *** zhill has joined #openstack-swift | 17:24 | |
| *** MooingLemur has joined #openstack-swift | 17:25 | |
| *** geaaru has quit IRC | 17:27 | |
| *** marcusvrn has quit IRC | 17:34 | |
| *** joeljwright has joined #openstack-swift | 17:48 | |
| minwoob | Have there been efforts to optimize the way that data is moved around, within the cluster? (e.g., during a rebalance, or replication activity, EC frag placement, etc) | 17:49 |
| notmyname | minwoob: yes, to some extent | 17:49 |
| notmyname | but most of the "optimization" is to not do it right now | 17:49 |
| minwoob | It seems that these are some of the operations that place the heaviest burden on the network. | 17:49 |
| notmyname | ie if an object is on server A but should be on server B, there isn't much we can do except move it from A to B | 17:50 |
| notmyname | torgomatic is working right now on some of the ring placement stuff where a ring rebalance may cause more partitions to move than necessary | 17:50 |
| notmyname | and kota_ implemented a patch a few months ago so that data is only sent to a remote region once per replication run on a given box | 17:51 |
| notmyname | that's the sort of things we can do | 17:52 |
| notmyname | minwoob: what are you thinking about? | 17:52 |
| *** lpabon has joined #openstack-swift | 18:02 | |
| *** aix has quit IRC | 18:09 | |
| minwoob | notmyname: The example you brought up in regards to torgomatic's optimization was more along the lines of what I was curious about. Basically if there were any discussions regarding how to make the movement of data within the network much smoother. | 18:09 |
| minwoob | notmyname: Moreso on a conceptual level right now, than something needing to be implemented right away. | 18:10 |
| minwoob | (just got back from an ad hoc meeting). | 18:11 |
| torgomatic | define "smoother" | 18:12 |
| torgomatic | (not trying to be confrontational, just not sure what that means in this context) | 18:12 |
| minwoob | torgomatic: smoother in terms of performance. | 18:13 |
| torgomatic | minwoob: moving fewer total partitions when rings change? moving fewer per rebalance but maybe having more rebalances? faster convergence of data after a rebalance? something else? | 18:14 |
| torgomatic | "performance" is a very broad term. | 18:14 |
| minwoob | torgomatic: I see what you're getting at. | 18:16 |
| minwoob | torgomatic: It looks like the moving fewer per rebalance but more rebalances scenario is more of a tradeoff, but the others seem like good criteria for "better performance". | 18:18 |
| minwoob | Maybe one to add there is how to relieve network congestion. | 18:19 |
| minwoob | Which would be a subset of some of the previously mentioned criteria. | 18:19 |
| minwoob | torgomatic: Regarding the convergence of data after a rebalance, are you referring to something other than the shifting of the partitions across the network? | 18:23 |
| torgomatic | minwoob: no, that's what I'm talking about. just that the data has moved to where the new ring says it should be | 18:24 |
| minwoob | I see. | 18:24 |
| *** NM has joined #openstack-swift | 18:29 | |
| *** mmcardle has joined #openstack-swift | 18:42 | |
| *** wbhuber has joined #openstack-swift | 18:49 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Prompt for missing username and/or password https://review.openstack.org/189387 | 18:52 |
| *** mmcardle has quit IRC | 18:54 | |
| notmyname | ok, we now have a slight restructuring of core reviewers in swift | 18:58 |
| notmyname | the group "swiftclient-core" now exists, and it is currently made up of "swift-core" and joeljwright | 18:58 |
| *** lpabon has quit IRC | 19:01 | |
| notmyname | joeljwright: thanks for helping out on the client work! | 19:02 |
| *** silor has quit IRC | 19:03 | |
| *** joeljwright has quit IRC | 19:10 | |
| *** joeljwright has joined #openstack-swift | 19:10 | |
| *** ChanServ sets mode: +v joeljwright | 19:12 | |
| tdasilva | joeljwright: welcome and great job! | 19:13 |
| *** rcoleman has joined #openstack-swift | 19:16 | |
| *** amoturi has quit IRC | 19:26 | |
| *** adutta has quit IRC | 19:31 | |
| openstackgerrit | Tim Burke proposed openstack/python-swiftclient: Add some bash helpers for auth stuff https://review.openstack.org/86224 | 19:42 |
| *** thurloat is now known as thurloat_isgone | 19:44 | |
| jrichli | joeljwright: congrats! | 19:46 |
| *** AbyssOne has quit IRC | 19:47 | |
| *** AbyssOne has joined #openstack-swift | 20:05 | |
| *** ronenkat has quit IRC | 20:16 | |
| *** acampbell has quit IRC | 20:19 | |
| *** takotuesday has joined #openstack-swift | 20:27 | |
| takotuesday | could there be any reason for sharepoint getting a 401 when trying to index a swift container that has ACLs set to global | 20:28 |
| *** fthiagogv_ has quit IRC | 20:34 | |
| *** fthiagogv has quit IRC | 20:34 | |
| *** fthiagogv has joined #openstack-swift | 20:39 | |
| *** fthiagogv_ has joined #openstack-swift | 20:39 | |
| notmyname | takotuesday: perhaps. maybe there's a large object manifest in there that's pointing to a different container and the ACLs aren't right on the other contianer | 20:42 |
| swifterdarrell | torgomatic: clayg: notmyname: anyone else: does anyone have a link handy to a good reference for all the crap you can put in docstrings for Sphinx auto-generated code stuff? I found this: http://sphinx-doc.org/domains.html#info-field-lists is that it? | 20:44 |
| torgomatic | swifterdarrell: I got nothin | 20:44 |
| MooingLemur | torgomatic: hiya... did you get a chance to backread from my mention last week about the non-healing EC object? I was afk from thursday afternoon onward, but I don't think I saw any mention from you (or at least no highlight) | 20:45 |
| takotuesday | notmyname: doesnt make sense though, it doesnt reference another container, its just full of static html files | 20:47 |
| takotuesday | .r:* are the acls | 20:47 |
| takotuesday | this is the sharepoint error: Access is denied. Verify that either the Default Content Access Account has access to this repository, or add a crawl rule to crawl this repository. If the repository being crawled is a SharePoint repository, verify that the account you are using has "Full Read" permissions on the SharePoint Web Application being crawled. | 20:48 |
| notmyname | none of the objects in the container are dynamic or static manifest objects? | 20:48 |
| notmyname | if not, then perhaps the auth token expired | 20:48 |
| torgomatic | MooingLemur: I took a brief look and didn't see any duplicate fragments; I have no idea why it's not working beyond that, though | 20:49 |
| torgomatic | there is a patch in Gerrit to fix something in the reconciler, but I'm not sure what | 20:49 |
| takotuesday | notmyname: none are manifest obj or dynamic | 20:49 |
| MooingLemur | object-reconstructor is leaking though too. After about 5 days, it's consuming about 20 gigs on each of the storage nodes. | 20:50 |
| MooingLemur | heh | 20:50 |
| openstackgerrit | Samuel Merritt proposed openstack/swift: Use just IP, not port, when determining partition placement https://review.openstack.org/191970 | 20:50 |
| torgomatic | swifterdarrell: ^^ | 20:50 |
| torgomatic | one line of real code change, plus docstrings and tests | 20:51 |
| *** wbhuber has quit IRC | 20:52 | |
| *** blmartin has quit IRC | 20:54 | |
| *** blmartin has joined #openstack-swift | 20:55 | |
| *** wbhuber has joined #openstack-swift | 20:57 | |
| *** amoturi has joined #openstack-swift | 21:01 | |
| *** petertr7 is now known as petertr7_away | 21:11 | |
| *** tellesnobrega has quit IRC | 21:11 | |
| clayg | torgomatic: yeah that's *surprisingly* stright forward? | 21:16 |
| *** cpen has joined #openstack-swift | 21:25 | |
| *** tellesnobrega has joined #openstack-swift | 21:39 | |
| *** fthiagogv has quit IRC | 21:40 | |
| *** fthiagogv_ has quit IRC | 21:40 | |
| *** AbyssOne has quit IRC | 21:48 | |
| *** ianbrown has joined #openstack-swift | 21:48 | |
| *** a1|away has quit IRC | 21:50 | |
| *** takotuesday has quit IRC | 21:52 | |
| *** jrichli has quit IRC | 21:56 | |
| *** chlong has quit IRC | 21:57 | |
| *** minwoob has quit IRC | 21:58 | |
| *** charlesw has joined #openstack-swift | 21:59 | |
| *** AbyssOne has joined #openstack-swift | 22:02 | |
| *** a1|away has joined #openstack-swift | 22:04 | |
| charlesw | HI folks, I have a two-region swift cluster. I wonder how to set up region_name in keystone middle config in my proxy-server.conf. I looked around to see if auth_region_name is supported but all I could find is https://bugs.launchpad.net/keystonemiddleware/+bug/1405717. Does anybody know if auth_region_name is supported? | 22:04 |
| openstack | Launchpad bug 1405717 in keystonemiddleware "region_name is not in keystone client auth_token config" [Wishlist,Confirmed] | 22:05 |
| *** NM has quit IRC | 22:05 | |
| *** blmartin has quit IRC | 22:10 | |
| mattoliverau | Morning | 22:17 |
| joeljwright | morning | 22:17 |
| joeljwright | mattoliverau: glad I spotted you, been needing to ping you about the technical writer you mentioned in Vancouver | 22:18 |
| mattoliverau | joeljwright: hey man, sure, when she comes online I'll introduce, unless your ending for the day, in which case I'll ping you her details in a private chat :) | 22:20 |
| joeljwright | yeah, sadly it's 23:20 here, so I think our only option is email | 22:20 |
| joeljwright | I've missed you by moments a few times now | 22:21 |
| joeljwright | damn these timezones! | 22:21 |
| notmyname | mattoliverau: me too! I wanted to ask about that too :-) | 22:21 |
| mattoliverau | joeljwright: lol, damn circular earth :P | 22:22 |
| joeljwright | source of all hassle: unicode and timezones | 22:22 |
| mattoliverau | notmyname: lol, I'll make sure your in the loop too :) | 22:22 |
| mattoliverau | joeljwright: its a double edged sword, 24 hour channel coverage.. but 24 hour channel coverage :P | 22:23 |
| joeljwright | :D | 22:23 |
| mattoliverau | Unless your a machine like clayg who during EC decided he just wouldn't sleep or notmyname who must nly sleep 3 hours a night :P | 22:24 |
| * mattoliverau goes to find ome coffee, cause its a cold and very rainy day here and so I still want to be in bed :P | 22:25 | |
| *** wbhuber has quit IRC | 22:28 | |
| *** david-lyle has quit IRC | 22:31 | |
| *** lcurtis has quit IRC | 22:36 | |
| *** charlesw has quit IRC | 22:44 | |
| *** cpen has quit IRC | 22:50 | |
| *** km has joined #openstack-swift | 22:57 | |
| *** mfalatic has joined #openstack-swift | 23:11 | |
| *** david-lyle has joined #openstack-swift | 23:11 | |
| *** asettle has quit IRC | 23:13 | |
| *** proteusguy has quit IRC | 23:16 | |
| *** kei_yama has joined #openstack-swift | 23:21 | |
| *** EmilienM|afk is now known as EmilienM | 23:24 | |
| openstackgerrit | Darrell Bishop proposed openstack/swift: Allow 1+ object-servers-per-disk deployment https://review.openstack.org/184189 | 23:26 |
| *** barker has joined #openstack-swift | 23:27 | |
| *** barker has quit IRC | 23:27 | |
| swifterdarrell | torgomatic: what's the right way to rebase my patch off https://review.openstack.org/#/c/191970/1 ? | 23:33 |
| swifterdarrell | torgomatic: I tried "git rebase bdb05abe4fcd93350bfdcf69a228cbaa576f4d8c" but that didn't work, even after some fetching | 23:34 |
| torgomatic | swifterdarrell: try `git review -d 191970` before rebasing; that'll get the commit pulled | 23:34 |
| swifterdarrell | torgomatic: do I do the gerrit "checkout" for your patch: git fetch https://review.openstack.org/openstack/swift refs/changes/70/191970/1 && git checkout FETCH_HEAD | 23:34 |
| swifterdarrell | torgomatic: then rebase off teh sha? | 23:34 |
| torgomatic | alternately, that fetch should do it | 23:34 |
| swifterdarrell | torgomatic: cool | 23:35 |
| torgomatic | you don't even need to check out that sha, just have it somewhere in your local repository | 23:35 |
| swifterdarrell | torgomatic: *nod*; and either of those cmds will do that, ya? | 23:35 |
| swifterdarrell | torgomatic: thx, i'm off the the races | 23:36 |
| torgomatic | swifterdarrell: cool | 23:36 |
| *** ho has joined #openstack-swift | 23:38 | |
| ho | good morning! | 23:43 |
| *** chlong has joined #openstack-swift | 23:47 | |
| *** barker has joined #openstack-swift | 23:47 | |
| *** barker has quit IRC | 23:50 | |
| *** zaitcev has joined #openstack-swift | 23:56 | |
| *** ChanServ sets mode: +v zaitcev | 23:56 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!