| openstackgerrit | Yogesh proposed openstack/manila master: Improve test coverage for share migration https://review.openstack.org/418559 | 00:31 |
|---|---|---|
| *** mtanino has quit IRC | 00:36 | |
| tommylikehu | hey bswartz are you still around? | 00:55 |
| tommylikehu | Just saw the new items you added on topic IPv6 on the PTG's etherpad :) | 01:01 |
| *** ganso has quit IRC | 01:04 | |
| *** gcb has joined #openstack-manila | 01:18 | |
| *** a-pugachev has quit IRC | 01:22 | |
| bswartz | tommylikehu: yeah I'm here | 01:30 |
| bswartz | tommylikehu: I read the scrollback this morning and saw your conversation with vponomaryov | 01:30 |
| bswartz | tommylikehu: we've discussed that topic several times and never really agreed on an answer | 01:31 |
| bswartz | tommylikehu: personally I'm leaning towards the 3rd option | 01:32 |
| bswartz | tbarron: ping | 01:32 |
| tommylikehu | bswartz: third option? | 01:32 |
| bswartz | tommylikehu: I think it makes sense for manila to always accept ip access rules in either format -- ipv4 or ipv6 | 01:33 |
| tommylikehu | Always allow rules even when not supported? | 01:33 |
| bswartz | backends can ignore unsupported ones but they should still get added to the manila DB | 01:33 |
| bswartz | the reason is because support for ipv6 could get added later | 01:34 |
| tommylikehu | that is reasonable, but even with that, we could only say that we support IPv6 in LVM and access with NAS, the rest part is never tested until the other drivers have supportted that. | 01:38 |
| bswartz | tommylikehu: I think that "ipvX_support" should mean: the driver code supports it, the backend code/hardware supports it, AND it's been administratively enabled on the backend | 01:39 |
| bswartz | if any one of those things is not true, then support should be false | 01:40 |
| tommylikehu | bswartz: then the 'ipv4_support' capabilities no longer belongs to the drivers only. | 01:41 |
| tommylikehu | ipvx_support | 01:41 |
| bswartz | well all drivers have code to support ipv4 today, and all backends support ipv4 today | 01:41 |
| bswartz | the only way ipv4 could not be supported, is if the backend had not been assigned an ipv4 address, or ipv4 had been administratively disabled | 01:42 |
| tommylikehu | yes, so you are leaning towards to vponomaryov's opinion | 01:43 |
| tommylikehu | bswartz I agree with you, the only thing that we should notice is that the we might update the manila's code when new driver which support new access type is going to support IPv6, cause that process has not been tested. | 01:49 |
| bswartz | tommylikehu: I'm more interested in situation where the backend gets reconfigured while Manila is running | 01:50 |
| bswartz | consider an existing share with 1 IPv4 export location, and then the administrator add and IPv6 address to the backend | 01:50 |
| tommylikehu | bswartz: restart the service | 01:51 |
| bswartz | how can manila discover the new IPv6 export location and ensure the access rules through that export location are correct? | 01:51 |
| tommylikehu | cinder guys is working on the dynammic reconfiguration | 01:51 |
| bswartz | I agree a service restart probably makes sense in that case, but we still lack the code to go update the shares that need updating in that situation | 01:51 |
| tommylikehu | bswartz: do we have the logic to update the share's export location when administrator add another IPv4 address? | 01:57 |
| bswartz | tommylikehu: no -- we had an agenda item to discuss that in Barcelona but we ran out of time | 01:59 |
| bswartz | we'll get to it in Atlanta for sure -- I need to spend some time thinking about it before then | 01:59 |
| tommylikehu | bswartz: so that might be another issue, but could be handled together with IPv6 | 02:00 |
| bswartz | there are lots of things that need to be checked and updated after a share is created | 02:00 |
| bswartz | the challenge is finding a way to do it that scales | 02:01 |
| bswartz | simply iterating over every share and updating it when the backend starts up will definitely not scale | 02:01 |
| tommylikehu | before that I am not sure whether this situation you mentioned is common for administrator , any statistics on that? | 02:02 |
| tommylikehu | especially when changing IPv4 to IPv6, that could make some access rule illegal for shares, | 02:03 |
| tommylikehu | administrator should realise that | 02:03 |
| bswartz | technically the rules aren't illegal -- they'll just never get used | 02:06 |
| bswartz | if I add an access rule to my share for ::/0 and the share only has IPv4 export locations, the access rule won't do anything | 02:06 |
| bswartz | IMO Manila should still accept and store that rule, because later the share may get an IPv6 export location and then the rule will make sense | 02:07 |
| tommylikehu | bswartz: if the case above is about adding or changing address with different ip version for same backend. the share should work after new access rule is added, and we do not need to update the shares. | 02:18 |
| tommylikehu | bswartz: and the only thing we need to do is dynamic build the export locations, correct? | 02:25 |
| *** crushil has joined #openstack-manila | 02:29 | |
| bswartz | tommylikehu: it's a matter of noticing when the the list of export locations for an existing share might have changed and then updating them all in the database | 02:40 |
| bswartz | I'm not sure how we can do it in a scalable way but it's something to think about | 02:41 |
| tommylikehu | sure, you are absolutely correct. maybe we can ask driver to update this when address is detected changing. | 02:43 |
| openstackgerrit | gecong proposed openstack/manila master: Replaces yaml.load() with yaml.safe_load() https://review.openstack.org/429156 | 02:44 |
| bswartz | gouthamr: regarding https://review.openstack.org/#/c/428398/6/manila/share/manager.py I really dislike the method name "get_and_UPDATE_share_instance_access_rules" when I only want to get the rules | 02:55 |
| gouthamr | bswartz: i know.. what else can we call that method? | 02:55 |
| gouthamr | bswartz: there're not a lot of places where you only get the rules... but you still want to hold a lock when doing so. | 02:56 |
| bswartz | gouthamr: maybe we need another method that wraps that method, removing all the upgrade functionality | 02:59 |
| bswartz | s/upgrade/update/ | 02:59 |
| gouthamr | bswartz: maybe.. but you could use that and set updates={}... wouldn't that solve the readability problem | 03:00 |
| bswartz | any time you want 2 different functionalities with 1 implementation you either wrap one method with the other or have both methods wrap a common private method | 03:01 |
| gouthamr | bswartz: that can work, but the common code is only two lines.. maybe it's worth just adding another method that takes the same named lock | 03:05 |
| bswartz | gouthamr: nobody ever needs to do both things at the same time? | 03:06 |
| gouthamr | bswartz: both things: get and update? most places need that.. | 03:06 |
| bswartz | well the splitting the method in 2 doesn't make sense | 03:07 |
| gouthamr | yes, i considered that deeply, i promise :) | 03:07 |
| gouthamr | i saw we only ever read the rules without updating once/twice | 03:08 |
| openstackgerrit | Merged openstack/manila master: Fix nonsense variable name https://review.openstack.org/428707 | 03:40 |
| *** ganso has joined #openstack-manila | 03:54 | |
| *** furlongm has quit IRC | 03:56 | |
| *** furlongm has joined #openstack-manila | 03:57 | |
| *** furlongm has quit IRC | 03:58 | |
| *** furlongm has joined #openstack-manila | 03:59 | |
| *** zhonghua has quit IRC | 04:11 | |
| *** zhonghua has joined #openstack-manila | 04:12 | |
| openstackgerrit | Yogesh proposed openstack/manila master: Improve test coverage for share migration https://review.openstack.org/418559 | 04:28 |
| *** crushil has quit IRC | 04:33 | |
| *** lpetrut has joined #openstack-manila | 04:49 | |
| *** harlowja has quit IRC | 05:08 | |
| *** gouthamr has quit IRC | 05:14 | |
| *** lpetrut has quit IRC | 05:16 | |
| *** lpetrut has joined #openstack-manila | 05:36 | |
| *** lpetrut has quit IRC | 05:40 | |
| *** ganso has quit IRC | 06:04 | |
| *** harlowja has joined #openstack-manila | 06:17 | |
| openstackgerrit | TommyLike proposed openstack/manila master: Enable IPv6 in manila(network plugins and drivers) https://review.openstack.org/406776 | 06:26 |
| *** crushil has joined #openstack-manila | 06:43 | |
| *** harlowja has quit IRC | 06:50 | |
| *** shausy has joined #openstack-manila | 06:58 | |
| *** shausy has quit IRC | 07:03 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/manila-ui master: Imported Translations from Zanata https://review.openstack.org/429194 | 07:19 |
| *** harlowja has joined #openstack-manila | 07:50 | |
| *** crushil has quit IRC | 07:53 | |
| *** harlowja has quit IRC | 07:57 | |
| tommylikehu | ping vponomaryov | 08:58 |
| tommylikehu | hey vponomaryov, I still reckon we do not need to use 'dummy_driver_dhss_false_ip_version' for both dhss false and true, I pasted the configuration on the four cases at the latest comments here: https://review.openstack.org/#/c/406776, please take a look and correct me if something is missing. | 09:12 |
| *** a-pugachev has joined #openstack-manila | 09:45 | |
| *** a-pugachev has quit IRC | 10:02 | |
| *** lpetrut has joined #openstack-manila | 10:06 | |
| tbarron | rhagarty: you need 39d6b1e12410f1edc9e80b5b8006fa7539771c4b for manila ui so that it doesn't call share_network_create with nova net argument (microversion 2.26 bump) | 10:16 |
| tbarron | bswartz: pong - saw your note on ppa if that was the ping, very cool, will try it out when I circle back to v6 testing | 10:38 |
| *** zhonghua has quit IRC | 10:46 | |
| *** zhonghua has joined #openstack-manila | 10:47 | |
| *** lpetrut has quit IRC | 10:51 | |
| openstackgerrit | Merged openstack/manila master: Prepare for using standard python tests https://review.openstack.org/429032 | 11:15 |
| *** kberger has quit IRC | 12:16 | |
| *** gcb has quit IRC | 12:34 | |
| *** dgonzalez_ has joined #openstack-manila | 12:53 | |
| *** dgonzalez_ has quit IRC | 12:55 | |
| *** xinyanzhang has quit IRC | 14:10 | |
| *** xinyanzhang has joined #openstack-manila | 14:11 | |
| *** gouthamr has joined #openstack-manila | 14:46 | |
| *** crushil has joined #openstack-manila | 15:26 | |
| *** jcsp has quit IRC | 16:03 | |
| *** kberger has joined #openstack-manila | 16:16 | |
| *** chlong has quit IRC | 16:17 | |
| *** gouthamr has quit IRC | 16:45 | |
| *** crushil has quit IRC | 16:56 | |
| *** crushil has joined #openstack-manila | 17:07 | |
| *** crushil has quit IRC | 18:00 | |
| *** kberger has quit IRC | 18:32 | |
| *** kberger has joined #openstack-manila | 18:32 | |
| *** gouthamr has joined #openstack-manila | 19:47 | |
| *** david-lyle_ has joined #openstack-manila | 21:02 | |
| *** kberger has quit IRC | 21:04 | |
| *** david-lyle has quit IRC | 21:05 | |
| *** lpetrut has joined #openstack-manila | 21:07 | |
| *** eharney has quit IRC | 21:26 | |
| *** catintheroof has joined #openstack-manila | 22:05 | |
| *** breitz has quit IRC | 22:27 | |
| *** breitz has joined #openstack-manila | 22:28 | |
| *** akerr has joined #openstack-manila | 22:29 | |
| *** akerr has quit IRC | 22:29 | |
| *** gouthamr has quit IRC | 22:51 | |
| *** david-lyle_ has quit IRC | 23:05 | |
| *** lpetrut has quit IRC | 23:06 | |
| *** david-lyle has joined #openstack-manila | 23:07 | |
| *** david-lyle has quit IRC | 23:11 | |
| *** catintheroof has quit IRC | 23:42 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!