| *** gyee has quit IRC | 01:45 | |
| *** psachin has joined #openstack-swift | 03:35 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: Skip container-sync when versioning enabled https://review.opendev.org/698139 | 06:14 |
|---|---|---|
| *** pcaruana has joined #openstack-swift | 06:18 | |
| *** gkadam has joined #openstack-swift | 07:13 | |
| *** gkadam has quit IRC | 07:13 | |
| *** tesseract has joined #openstack-swift | 07:59 | |
| *** tkajinam has quit IRC | 08:04 | |
| *** rdejoux has joined #openstack-swift | 08:33 | |
| *** corvus has quit IRC | 08:40 | |
| *** tonyb has quit IRC | 08:56 | |
| *** tonyb has joined #openstack-swift | 08:56 | |
| *** corvus has joined #openstack-swift | 08:59 | |
| *** rpittau|afk is now known as rpittau | 09:00 | |
| *** ccamacho has joined #openstack-swift | 09:35 | |
| *** ccamacho is now known as ccamacho|pto | 09:48 | |
| *** mikecmpbll has joined #openstack-swift | 10:27 | |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: Skip container-sync when versioning enabled https://review.opendev.org/698139 | 10:45 |
| *** rcernin has quit IRC | 10:56 | |
| *** pcaruana has quit IRC | 11:03 | |
| *** pcaruana has joined #openstack-swift | 11:06 | |
| *** FlorianFa has joined #openstack-swift | 13:22 | |
| *** pcaruana has quit IRC | 13:57 | |
| henriqueof | Soppose my controllers are down, is there a way to recover my backups from a wsift node? | 14:32 |
| *** pcaruana has joined #openstack-swift | 14:36 | |
| DHE | not unless it's marked as a public container (sounds unlikely), or modifying a proxy-server to remove authentication requirements (not recommended!) | 14:51 |
| DHE | keystone should be redundant if that's a concern | 14:51 |
| *** ccamacho|pto has quit IRC | 14:54 | |
| *** henriqueof has quit IRC | 15:08 | |
| *** henriqueof1 has joined #openstack-swift | 15:08 | |
| clayg | sorry i missed the meeting yesterday, just read logs | 15:35 |
| *** gyee has joined #openstack-swift | 16:05 | |
| *** irclogbot_2 has quit IRC | 16:14 | |
| *** irclogbot_1 has joined #openstack-swift | 16:16 | |
| *** tesseract has quit IRC | 16:48 | |
| *** rdejoux has quit IRC | 16:56 | |
| *** efried is now known as efried_afk | 17:08 | |
| *** mikecmpbll has quit IRC | 17:39 | |
| *** rpittau is now known as rpittau|afk | 17:49 | |
| *** psachin has quit IRC | 18:09 | |
| *** gmann is now known as gmann_afk | 18:53 | |
| *** efried_afk is now known as efried | 19:38 | |
| *** pcaruana has quit IRC | 20:19 | |
| openstackgerrit | Tim Burke proposed openstack/swift master: Add normalize_etag() helper function https://review.opendev.org/698799 | 20:32 |
| openstackgerrit | Tim Burke proposed openstack/swift master: Allow clients to send quoted ETags for static links https://review.opendev.org/698800 | 20:32 |
| clayg | timburke: ^ and those are orthogonal to the new proxy option? | 20:36 |
| clayg | oh oh, i see - you're gunna rebase 👍 | 20:37 |
| timburke | yup! just breaking them out as that option seems like it may instead become a per-container setting implemented in middleware | 20:37 |
| timburke | yeah, maybe rebase, maybe abandon -- either way, those seem like independent changes that could merge on their own merits | 20:37 |
| DHE | how bad is this action? cd /srv/node ; for i in slot*; do swift-object-replicator -v -o -d $i /etc/swift/object-server.conf & sleep 1 ; done | 20:57 |
| DHE | background: a region with only 2 hosts (most disks on a single host) was expanded to 5 hosts. this command was run on that "most disks" host | 20:59 |
| timburke | running replicators is rarely "bad" | 21:00 |
| timburke | makes me wonder why we're running it as a once-off thing, though. and why we don't just set replicator_workers to the number of disks | 21:01 |
| timburke | is there any bad behavior you're observing that's prompting the question? | 21:02 |
| DHE | I set concurrency=4, workers=4 so maybe that needs to be cranked | 21:03 |
| DHE | number of disks is 30 | 21:03 |
| DHE | running as a one-off just to get the partitions redistributed quickly | 21:04 |
| timburke | i could see logs getting noisy or disks getting overloaded if you've got multiple replicators trying to process the same disk at the same time... but shouldn't be anything that'd put durability at risk, say | 21:04 |
| timburke | what's your replication cycle time like? | 21:05 |
| DHE | defaults | 21:05 |
| DHE | this is my first big replication event and I felt like it was underperforming for what should have been a pretty easy port saturation... | 21:07 |
| openstackgerrit | Thiago da Silva proposed openstack/swift master: Skip container-sync when versioning enabled https://review.opendev.org/698139 | 21:25 |
| openstackgerrit | Clay Gerrard proposed openstack/python-swiftclient master: object versioning features https://review.opendev.org/691877 | 21:35 |
| clayg | DHE: workers == # disks with a concurrency of 2-4 is normally enough to get the job done - it helps if the receiving end has lots of head room to accept rsync connections | 21:39 |
| clayg | if this is your "first big replication event" - it's totally normal to be looking at network graphs and disk utilization and doing some tuning/optimization of settings | 21:40 |
| clayg | also reasonable to stop/slowdown the background replicators and play with settings using smaller targeted runs until you get a feel for what you want to do | 21:41 |
| clayg | i.e. your for loop is very good and normal way to be thinking about the replicators | 21:41 |
| clayg | the trick is to get a sense for what your cluster can do and find some settings you can run with so you don't have to monkey around with every rebalance | 21:42 |
| clayg | although TBH - we very often turn on handoffs_only after a big ring change to get things moving | 21:42 |
| *** dosaboy has quit IRC | 22:01 | |
| *** aluria has quit IRC | 22:02 | |
| *** henriqueof1 has quit IRC | 22:08 | |
| mattoliverau | morning | 22:13 |
| tdasilva | mattoliverau: o/ | 22:15 |
| *** gmann_afk is now known as gmann | 22:29 | |
| openstackgerrit | Clay Gerrard proposed openstack/swift master: WSGI server workers must drop_privledges https://review.opendev.org/698563 | 23:02 |
| *** tkajinam has joined #openstack-swift | 23:06 | |
| *** irclogbot_1 has quit IRC | 23:11 | |
| *** irclogbot_2 has joined #openstack-swift | 23:13 | |
| *** dosaboy has joined #openstack-swift | 23:23 | |
| *** diablo_rojo has quit IRC | 23:24 | |
| *** dosaboy has quit IRC | 23:29 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!