Sunday, 2016-05-08

*** x00350071 has quit IRC06:08
*** x00350071 has joined #openstack-smaug06:09
*** yuval has joined #openstack-smaug06:51
*** gampel has joined #openstack-smaug07:30
*** c00281451_ has joined #openstack-smaug07:38
*** c00281451 has quit IRC07:38
openstackgerritYuval Brik proposed openstack/smaug: BankSection: implement get sub-section  https://review.openstack.org/31161407:38
openstackgerritYuval Brik proposed openstack/smaug: BankSection: implement get sub-section  https://review.openstack.org/31161410:29
*** saggi has joined #openstack-smaug10:50
saggiyuval: Please review ( https://review.openstack.org/#/c/229722/17/specs/newton/replication-group.rst )10:53
yuvalsaggi: thanks10:54
*** gampel has quit IRC11:08
*** gampel has joined #openstack-smaug11:08
*** yuval_ has joined #openstack-smaug11:08
*** yuval has quit IRC11:08
*** saggi has quit IRC11:09
*** saggi has joined #openstack-smaug11:10
saggiyuval: Cinder's weekly meeting is Wed at 19:00 IST11:14
*** yuval_ is now known as yuval11:33
saggiyuval: wrt your question about why delete marks for deletion and than deletion happens in the background instead of actually deleting.11:48
saggiany operation can fail, especially a complex operation like deletion11:49
saggiBut the user would like it to always succeed.11:49
saggiAlso, this is to prevent collision between multiple sites.11:50
saggiSince this doesn't require a lease.11:50
yuvalWriting to 'indice' bank section does require a lease.11:51
sagginot for deletion.11:52
saggiwell not exactly11:53
yuvaland when the GC actually removes the checkpoint, does it remove it from the 'delete checkpoints' index?11:53
saggiIf a lease exists I fail11:53
saggiBut I don't need to acquire one11:53
yuval(that means that delete can fail, if a lease exists)11:53
saggiSo this means that if there is no lease on a checkpoint. Anyone can just mark it for deletion. This means that after a certain amount of time I can be assured that no one is restoring from it.11:54
saggiYes, but immediately. Not because of some arcane storage problem. And it will never be in a half state.11:54
saggiIf it fails during actual deletion (or purge) than we the admin will have to fix it. But the checkpoint has been invalid for restore for a long time.11:55
saggiThe main issue is if we are restoring on one site and deleting on another.11:55
yuvalI see11:55
saggiSo even if you marked for deletion. The restore can succeed since no data is being deleted,11:56
saggiYou just can't start new restores.11:56
saggiOnce we know all sites know about the deletion we can start deleting11:56
saggiThere might be a collision during purge11:56
saggibut this is OK11:56
saggisince we can just try again later.11:56
saggihopefully it will converge.11:57
saggiBut it's not something that the tenant cares about.11:57
saggi(except for his qoutas still being used).11:57
yuvalIf the GC runs in a fixed interval, it might 'hit' the exact moment after someone marked a checkpoint for deletion, and start GC the data11:58
yuvalwe also might have an issue with multiple protection services whose GCs might collide12:00
saggiAs I said, we don't care if multiple GCs collide. The cleanup should handle that and just be OK if someone deleted the resource for you.12:04
saggiThe GC will need to only delete entities that have been marked a certain amount of time ago.12:04
saggiIt's all a bit flaky but we are trying to avoid having cross site distributed locks so we configure the maximum time for global replication and use that.12:04
saggiDLM over WAN is not a thing you can scale.12:05
saggiIt's also a very rare edge case so even if there are some places where it might break it12:07
saggiit's not that bad since we are talking about something that is being deleted.12:07
saggiYou really should only delete things you know you don't need.12:07
saggiWe also need the GC since cleanup might involve multiple sites (though we are not supporting this in V1).12:08
saggiSo all sites will clean up and when someone arrive to the conclusion that everything is free actually remove the index.12:08
saggiyuval: ^^^^12:12
yuvalsaggi: I see, thanks12:12
*** yuval has quit IRC12:29
*** yuval_ has joined #openstack-smaug12:29
*** gampel1 has joined #openstack-smaug12:29
*** saggi has quit IRC12:29
*** saggi has joined #openstack-smaug12:29
*** gampel has quit IRC12:30
*** yuval_ is now known as yuval12:34
openstackgerritYuval Brik proposed openstack/smaug: BankSection: implement get sub-section  https://review.openstack.org/31161414:26
openstackgerritYuval Brik proposed openstack/smaug: BankSection: implement get sub-section  https://review.openstack.org/31161414:34
*** yuval has quit IRC14:55
*** chenying has quit IRC15:29
*** chenying has joined #openstack-smaug15:30
*** gampel has joined #openstack-smaug15:51
*** zhonghua-lee has quit IRC16:06
*** zhonghua-lee has joined #openstack-smaug16:07
*** saggi has quit IRC16:09
*** gampel has quit IRC17:54
*** saggi has joined #openstack-smaug19:25
*** saggi has quit IRC21:11
*** gampel1 has quit IRC23:25
*** gampel has joined #openstack-smaug23:39

Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!