| *** janonymous has joined #openstack-swift | 00:10 | |
| *** tdasilva has quit IRC | 00:16 | |
| openstackgerrit | Matthew Treinish proposed openstack/swift: Use pbr console_script entrypoints for bin/ scripts https://review.openstack.org/416776 | 00:17 |
|---|---|---|
| openstackgerrit | Matthew Treinish proposed openstack/swift: Use pbr console_script entrypoints for bin/ scripts https://review.openstack.org/416776 | 00:30 |
| dims | mtreinish : ack thanks | 00:38 |
| mtreinish | dims: although timburke brought up a good point the scripts still get their shebang adjusted with setuptools in the current setup.cfg | 00:44 |
| mtreinish | the reason it's failing for us now with py35 is because lib/swift calls SWIFT_DIR/bin/* manually instead of the installed ones | 00:44 |
| *** manous has quit IRC | 00:55 | |
| *** jamielennox is now known as jamielennox|away | 01:02 | |
| *** winggundamth has joined #openstack-swift | 01:09 | |
| *** winggundamth has quit IRC | 01:17 | |
| *** jamielennox|away is now known as jamielennox | 01:21 | |
| *** winggundamth has joined #openstack-swift | 01:23 | |
| *** garyj has quit IRC | 01:24 | |
| *** furlongm has quit IRC | 01:36 | |
| *** furlongm has joined #openstack-swift | 01:36 | |
| *** bkopilov has quit IRC | 01:41 | |
| *** bkopilov has joined #openstack-swift | 01:52 | |
| * kota_ is back to online at his office | 01:59 | |
| *** klrmn has quit IRC | 02:05 | |
| *** vinurs has joined #openstack-swift | 02:06 | |
| *** JimCheung has left #openstack-swift | 02:08 | |
| *** tqtran has quit IRC | 02:31 | |
| *** trananhkma has quit IRC | 02:34 | |
| *** bkopilov has quit IRC | 03:04 | |
| *** tonyb has quit IRC | 03:05 | |
| *** treyd has quit IRC | 03:05 | |
| *** sgundur has quit IRC | 03:05 | |
| *** acoles_ has quit IRC | 03:05 | |
| *** timss has quit IRC | 03:05 | |
| *** mminesh has quit IRC | 03:05 | |
| *** charz has quit IRC | 03:05 | |
| *** verto has quit IRC | 03:05 | |
| *** andymccr has quit IRC | 03:05 | |
| *** lxsli has quit IRC | 03:05 | |
| *** mminesh` has joined #openstack-swift | 03:05 | |
| *** tonyb has joined #openstack-swift | 03:05 | |
| *** verto_ has joined #openstack-swift | 03:05 | |
| *** lxsli has joined #openstack-swift | 03:05 | |
| *** andymccr_ has joined #openstack-swift | 03:05 | |
| *** JimCheung has joined #openstack-swift | 03:05 | |
| *** timss has joined #openstack-swift | 03:05 | |
| *** mminesh` is now known as mminesh | 03:05 | |
| *** charz has joined #openstack-swift | 03:06 | |
| *** sgundur has joined #openstack-swift | 03:06 | |
| *** treyd has joined #openstack-swift | 03:06 | |
| *** acoles_ has joined #openstack-swift | 03:06 | |
| *** ChanServ sets mode: +v acoles_ | 03:06 | |
| *** acoles_ is now known as acoles | 03:06 | |
| *** JimCheung has quit IRC | 03:10 | |
| jrichli | anyone else only getting some of the email notifications for patches, but not all that they used to get? | 03:32 |
| *** klrmn has joined #openstack-swift | 03:37 | |
| *** links has joined #openstack-swift | 03:40 | |
| *** Jeffrey4l has quit IRC | 04:06 | |
| *** Jeffrey4l has joined #openstack-swift | 04:07 | |
| *** tqtran has joined #openstack-swift | 04:27 | |
| *** psachin has joined #openstack-swift | 04:29 | |
| *** tqtran has quit IRC | 04:30 | |
| *** bkopilov has joined #openstack-swift | 04:30 | |
| *** furlongm has quit IRC | 04:35 | |
| *** furlongm has joined #openstack-swift | 04:36 | |
| *** ppai has joined #openstack-swift | 05:10 | |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 05:15 |
| *** ksameersrk has joined #openstack-swift | 05:17 | |
| *** jamielennox is now known as jamielennox|away | 05:23 | |
| ksameersrk | Hi All, I was eager to know the data-flow of an object on Openstack Swift PUT command. I know a highlevel overview from various posts, which I am not sure of. Client Initiates PUT command specifying the object path on local storage to that on swift cloud. The Object is transferred to proxy-server over HTTP request. If (REPLICAS = 3), then three primary nodes are found out using Ring Algorithm. Then the object is transfer | 05:24 |
| ksameersrk | *from proxy server. Then after majority successful ACK (two here), the client is sent back ACK. I will be happy if someone can confirm the above sequence of steps is correct. My follow questions are What happens when there is NO ACK from the 3rd node? How is rsync used, when there is node failure? | 05:25 |
| *** tqtran has joined #openstack-swift | 05:30 | |
| *** jamielennox|away is now known as jamielennox | 05:30 | |
| *** Jeffrey4l_ has joined #openstack-swift | 05:31 | |
| *** tqtran has quit IRC | 05:32 | |
| *** Jeffrey4l has quit IRC | 05:34 | |
| *** dfflanders has quit IRC | 05:41 | |
| *** SkyRocknRoll has joined #openstack-swift | 05:46 | |
| *** ksameersrk has quit IRC | 06:11 | |
| *** chsc has joined #openstack-swift | 06:17 | |
| *** chsc has quit IRC | 06:21 | |
| *** McMurlock1 has joined #openstack-swift | 06:52 | |
| *** pcaruana has joined #openstack-swift | 07:05 | |
| *** tesseract has joined #openstack-swift | 07:09 | |
| *** ChubYann has quit IRC | 07:16 | |
| *** rcernin has joined #openstack-swift | 07:17 | |
| *** hseipp has joined #openstack-swift | 07:42 | |
| *** openstackgerrit has quit IRC | 07:50 | |
| *** zhugaoxiao has quit IRC | 07:54 | |
| *** zhugaoxiao has joined #openstack-swift | 07:54 | |
| donagh | ksameersrk: your sequence is mostly correct. Except, if the obj node is down before you start the transaction, the proxy will notice and send the object to yet another node (called the handoff). So in that case you end up with 3 replicas on disk. Your sequence is correct when the obj node dies in the middle of the transaction. | 07:59 |
| donagh | When the node is restored/rebooted/fixed, the object replicators will use rsync to copy the data from one of the other replicas. | 08:02 |
| donagh | This could be one of the primary copies or a handoff - does not matter .. the first to notice will fix up the missing replica | 08:03 |
| donagh | The handoff node will then delete its copy once it knows the originally-down node has a good copy | 08:03 |
| *** rledisez has joined #openstack-swift | 08:17 | |
| rledisez | hello kota_, are you around? | 08:20 |
| *** ianstanfield has quit IRC | 08:46 | |
| *** oshritf has joined #openstack-swift | 08:48 | |
| *** openstackgerrit has joined #openstack-swift | 08:53 | |
| openstackgerrit | Christian Schwede proposed openstack/swift: Add support to increase object ring partition power https://review.openstack.org/337297 | 08:53 |
| *** cbartz has joined #openstack-swift | 09:07 | |
| *** asettle_ has joined #openstack-swift | 09:08 | |
| *** cbartz has quit IRC | 09:08 | |
| *** andymccr_ is now known as andymccr | 09:10 | |
| *** abhitechie has joined #openstack-swift | 09:30 | |
| *** tqtran has joined #openstack-swift | 09:32 | |
| *** openstackgerrit has quit IRC | 09:33 | |
| *** svp has joined #openstack-swift | 09:33 | |
| *** tqtran has quit IRC | 09:33 | |
| *** svp is now known as Guest60001 | 09:34 | |
| Guest60001 | hi, i wanted to know the role of rsync in object replication | 09:35 |
| *** ianstanfield has joined #openstack-swift | 09:42 | |
| kota_ | rledisez: here | 09:47 |
| *** ianstanfield has quit IRC | 09:47 | |
| *** Guest60001 has quit IRC | 09:48 | |
| rledisez | kota_: hello :) | 09:49 |
| kota_ | rledisez: hello, hapy new year :-) | 09:49 |
| rledisez | thx, happy new year to you too :) | 09:49 |
| rledisez | in barcelona we had nice discussion about https://review.openstack.org/#/c/287057/ | 09:49 |
| patchbot_ | patch 287057 - swift - WIP: Swift Automated Tiering | 09:49 |
| rledisez | i think it was with tdasilva- | 09:50 |
| kota_ | rledisez: sure | 09:50 |
| rledisez | i see that the review did not evolved much since | 09:50 |
| rledisez | we are very interrested in this feature, so we would like to offer you help to make it alive again | 09:50 |
| rledisez | if that's ok with you | 09:50 |
| kota_ | ah, it looks like the last updated is... Oct!?!?!? | 09:51 |
| kota_ | i saw the main author, m_kazuhiro, for that pach was working recently | 09:52 |
| rledisez | yeah :) maybe you don't have time to work on this, or something is blocking you (we saw that few changes were needed about the devices scanning thing) | 09:52 |
| rledisez | oh, ok | 09:52 |
| kota_ | he sits down next door. I will poke him to update | 09:52 |
| rledisez | ok, thx :) | 09:52 |
| kota_ | i think he has progress with tdasilva, iirc, he had telconf with him the end of the last year. | 09:53 |
| *** mvk has quit IRC | 09:53 | |
| rledisez | ok. i would be interested to be in the loop if possible. i can have a colleague work on that for few weeks if needed. we really want this to be upstream ;) | 09:54 |
| kota_ | probably, m_kazhiro has diff since the patch set 50 and, if you prefer to review that patch, i think it's nice to happen after updating. | 09:54 |
| kota_ | sure | 09:54 |
| *** m_kazuhiro has joined #openstack-swift | 09:58 | |
| m_kazuhiro | good morning. | 09:58 |
| rledisez | m_kazuhiro: hello :) | 09:59 |
| m_kazuhiro | rledisez: hello! | 10:00 |
| rledisez | m_kazuhiro: as i was telling kota_, we have serious interest in having https://review.openstack.org/#/c/287057/ merged upstream | 10:01 |
| patchbot_ | patch 287057 - swift - WIP: Swift Automated Tiering | 10:01 |
| rledisez | m_kazuhiro: so, if we can help in any way, we will be happy to do so | 10:01 |
| rledisez | m_kazuhiro: I have a developper that can work almost full time on this for few weeks if it can help | 10:02 |
| *** sheel has joined #openstack-swift | 10:03 | |
| m_kazuhiro | rledisez: Thank you for your suggestion. In current status, code review on gerrit will help me. | 10:05 |
| m_kazuhiro | rledisez: There will be small changes on tiering patch for comments in the last summit. But I think we can start reviewing. | 10:08 |
| rledisez | m_kazuhiro: ok, i'll take time to see that when the review will be updated | 10:08 |
| *** geaaru has joined #openstack-swift | 10:09 | |
| m_kazuhiro | rledisez: Thank you. I will push local change on tiering patch to gerrit today. | 10:11 |
| *** openstackstatus has quit IRC | 10:13 | |
| *** openstack has joined #openstack-swift | 10:16 | |
| *** openstackstatus has joined #openstack-swift | 10:16 | |
| *** ChanServ sets mode: +v openstackstatus | 10:16 | |
| *** openstackgerrit has joined #openstack-swift | 10:20 | |
| openstackgerrit | Falk Reimann proposed openstack/swift: Preserve trailing slash in domain remap https://review.openstack.org/416924 | 10:20 |
| *** manous has joined #openstack-swift | 10:21 | |
| *** mvk has joined #openstack-swift | 10:25 | |
| m_kazuhiro | rledisez: Are you still here? | 10:36 |
| rledisez | m_kazuhiro: yes | 10:37 |
| m_kazuhiro | rledisez: I have an additional request on my tiering patch. | 10:38 |
| m_kazuhiro | rledisez: One of my largest task to merge automated tiering is making document. | 10:39 |
| m_kazuhiro | rledisez: But I'm not good at English, so the document task will be hard. Therefore, I want support to make the document. | 10:42 |
| rledisez | m_kazuhiro: sure. if you write a first draft, I can have a look at it to improve (my english is not perfect too, but I'll do my best ;)) | 10:45 |
| *** furlongm has quit IRC | 10:48 | |
| *** furlongm_ has joined #openstack-swift | 10:48 | |
| m_kazuhiro | rledisez: Thank you. I will make the first draft. | 10:51 |
| *** jordanP has joined #openstack-swift | 10:55 | |
| *** sams-gleb has joined #openstack-swift | 10:57 | |
| *** sams-gleb has joined #openstack-swift | 10:57 | |
| *** kei_yama has quit IRC | 11:12 | |
| *** furlongm has joined #openstack-swift | 11:12 | |
| *** furlongm_ has quit IRC | 11:13 | |
| *** nikivi has joined #openstack-swift | 11:15 | |
| *** abhitechie has quit IRC | 11:17 | |
| *** vint_bra has joined #openstack-swift | 11:17 | |
| *** nikivi has quit IRC | 11:20 | |
| *** vint_bra has quit IRC | 11:21 | |
| *** asettle_ has quit IRC | 11:22 | |
| *** asettle_ has joined #openstack-swift | 11:23 | |
| *** asettle_ is now known as asettle | 11:25 | |
| *** nikivi has joined #openstack-swift | 11:28 | |
| *** psachin has quit IRC | 11:43 | |
| *** ianstanfield has joined #openstack-swift | 11:43 | |
| *** timss has quit IRC | 11:44 | |
| *** psachin has joined #openstack-swift | 11:44 | |
| *** timss has joined #openstack-swift | 11:44 | |
| *** JimCheung has joined #openstack-swift | 11:45 | |
| *** ianstanfield has quit IRC | 11:48 | |
| *** svp has joined #openstack-swift | 11:49 | |
| *** svp is now known as Guest63092 | 11:49 | |
| Guest63092 | hi | 11:50 |
| *** JimCheung has quit IRC | 11:50 | |
| Guest63092 | hi | 11:51 |
| *** abhitechie has joined #openstack-swift | 11:52 | |
| Guest63092 | hi, i wanted to know the role of rsync in object replication | 11:52 |
| Guest63092 | once the proxy server creates a copy of the object on the 3 nodes(assuming replication factor = 3), where is rsync used? | 11:53 |
| Guest63092 | is it used if there is a failure of a node during copy or in other scenarios as well ? | 11:54 |
| *** Raymii has quit IRC | 11:55 | |
| *** glb has quit IRC | 11:55 | |
| *** mtreinish has quit IRC | 11:55 | |
| *** sileht has quit IRC | 11:55 | |
| *** htruta has quit IRC | 11:55 | |
| *** topol has quit IRC | 11:55 | |
| *** balajir has quit IRC | 11:55 | |
| *** mahatic_ has quit IRC | 11:55 | |
| *** m3m0_ has quit IRC | 11:55 | |
| *** dosaboy has quit IRC | 11:55 | |
| *** Anticimex has quit IRC | 11:55 | |
| *** dosaboy has joined #openstack-swift | 11:55 | |
| *** topol_ has joined #openstack-swift | 11:55 | |
| *** Raymii has joined #openstack-swift | 11:55 | |
| *** htruta has joined #openstack-swift | 11:55 | |
| *** mahatic has joined #openstack-swift | 11:56 | |
| *** glb has joined #openstack-swift | 11:56 | |
| *** mtreinish has joined #openstack-swift | 11:56 | |
| *** sileht has joined #openstack-swift | 11:56 | |
| *** balajir has joined #openstack-swift | 11:56 | |
| *** m3m0 has joined #openstack-swift | 11:58 | |
| *** serverascode has quit IRC | 11:59 | |
| *** Guest63092 has quit IRC | 11:59 | |
| *** serverascode has joined #openstack-swift | 12:01 | |
| *** nikivi has quit IRC | 12:04 | |
| *** nikivi has joined #openstack-swift | 12:06 | |
| *** links has quit IRC | 12:13 | |
| *** bkopilov has quit IRC | 12:20 | |
| *** links has joined #openstack-swift | 12:25 | |
| acoles | rsync is used whenever the replication process detects that replica is missing from a primary node. Replicas may be missing because the original put from proxy failed, or because a disk later failed, or because a disk error corrupted the file and the auditor quarantined the bad file. | 12:29 |
| openstackgerrit | Falk Reimann proposed openstack/swift: Preserve trailing slash in domain remap https://review.openstack.org/416924 | 12:33 |
| openstackgerrit | Kazuhiro MIYAHARA proposed openstack/swift: WIP: Swift Automated Tiering https://review.openstack.org/287057 | 12:36 |
| *** catintheroof has quit IRC | 12:38 | |
| *** m_kazuhiro has quit IRC | 12:42 | |
| *** catintheroof has joined #openstack-swift | 12:44 | |
| *** psachin has quit IRC | 12:48 | |
| *** vinsh_ has quit IRC | 12:51 | |
| *** psachin has joined #openstack-swift | 12:55 | |
| *** links has quit IRC | 12:55 | |
| *** catintheroof has quit IRC | 12:57 | |
| *** catintheroof has joined #openstack-swift | 12:58 | |
| *** catintheroof has quit IRC | 13:02 | |
| *** catintheroof has joined #openstack-swift | 13:04 | |
| *** links has joined #openstack-swift | 13:07 | |
| *** catintheroof has quit IRC | 13:11 | |
| *** Anticimex has joined #openstack-swift | 13:25 | |
| *** abhitechie has quit IRC | 13:27 | |
| *** d0ugal has quit IRC | 13:34 | |
| *** d0ugal has joined #openstack-swift | 13:35 | |
| *** klamath has joined #openstack-swift | 13:37 | |
| *** klamath has quit IRC | 13:37 | |
| *** klamath has joined #openstack-swift | 13:38 | |
| *** ianstanfield has joined #openstack-swift | 13:45 | |
| *** catintheroof has joined #openstack-swift | 13:47 | |
| *** nikivi has quit IRC | 13:48 | |
| *** ianstanfield has quit IRC | 13:50 | |
| *** cdelatte has joined #openstack-swift | 13:53 | |
| jrichli | mathiasb, mahatic, acoles: I see tdasilva- is away, and he hasn't accepted the invite for the upcoming video conf. | 13:56 |
| *** etienneme has joined #openstack-swift | 13:58 | |
| acoles | I can imagine he is just pulling into the parking lot :) | 13:58 |
| tdasilva- | good morning | 13:59 |
| mahatic | yay he is here | 13:59 |
| jrichli | oh, godd. tdasilva- o/ | 13:59 |
| tdasilva- | sorry i'm late | 13:59 |
| mahatic | tdasilva-: good morning | 13:59 |
| jrichli | s/godd/good/ | 13:59 |
| acoles | I made an etherpad to capture some thoughts https://etherpad.openstack.org/p/swift_keymaster | 13:59 |
| mahatic | tdasilva-: you're not late, we were looking out early ;) | 13:59 |
| tdasilva- | i'm in the meeting here: https://bluejeans.com/933935585 | 14:01 |
| acoles | me too | 14:01 |
| jrichli | im working on it | 14:02 |
| *** links has quit IRC | 14:02 | |
| jrichli | i thought this was already installed, but it seemed to think it had to install. launcher is taking awhile | 14:03 |
| acoles | I'm spinning on "Connecting" | 14:04 |
| mathiasb | acoles: we saw your chat | 14:04 |
| acoles | k | 14:04 |
| mathiasb | acoles: but now you've dropped | 14:06 |
| *** caiobrentano has quit IRC | 14:09 | |
| *** caiobrentano has joined #openstack-swift | 14:10 | |
| *** bkopilov has joined #openstack-swift | 14:10 | |
| *** vinurs has quit IRC | 14:11 | |
| *** vinurs has joined #openstack-swift | 14:11 | |
| *** vinurs has quit IRC | 14:12 | |
| *** vinurs has joined #openstack-swift | 14:12 | |
| *** caiobrentano has quit IRC | 14:14 | |
| *** caiobrentano has joined #openstack-swift | 14:16 | |
| *** vinurs has quit IRC | 14:17 | |
| *** _JZ_ has joined #openstack-swift | 14:18 | |
| *** Jeffrey4l_ has quit IRC | 14:21 | |
| openstackgerrit | Christian Schwede proposed openstack/swift: Add support to increase object ring partition power https://review.openstack.org/337297 | 14:23 |
| *** _JZ__ has joined #openstack-swift | 14:25 | |
| *** nikivi has joined #openstack-swift | 14:31 | |
| *** Jeffrey4l_ has joined #openstack-swift | 14:35 | |
| jrichli | we lost you acoles | 14:36 |
| *** wer has quit IRC | 14:42 | |
| *** wer has joined #openstack-swift | 14:43 | |
| *** vinurs has joined #openstack-swift | 14:45 | |
| *** Jeffrey4l_ has quit IRC | 14:52 | |
| *** chlong has joined #openstack-swift | 15:05 | |
| *** Jeffrey4l_ has joined #openstack-swift | 15:05 | |
| *** psachin has quit IRC | 15:06 | |
| *** sams-gleb has quit IRC | 15:15 | |
| *** cnf has quit IRC | 15:15 | |
| *** sams-gleb has joined #openstack-swift | 15:15 | |
| *** cnf has joined #openstack-swift | 15:16 | |
| *** sams-gleb has quit IRC | 15:20 | |
| openstackgerrit | Falk Reimann proposed openstack/swift: CNAME lookup within storage domain https://review.openstack.org/417010 | 15:23 |
| *** tdasilva- is now known as tdasilva | 15:23 | |
| *** caiobrentano has quit IRC | 15:23 | |
| *** ChanServ sets mode: +v tdasilva | 15:23 | |
| *** caiobrentano has joined #openstack-swift | 15:27 | |
| *** vint_bra has joined #openstack-swift | 15:28 | |
| tdasilva | acoles, jrichli, mathiasb, mahatic: added a link of the recording to the etherpad | 15:29 |
| jrichli | tdasilva: great, thanks! | 15:30 |
| *** sams-gleb has joined #openstack-swift | 15:32 | |
| *** catintheroof has quit IRC | 15:32 | |
| *** Jeffrey4l_ has quit IRC | 15:36 | |
| acoles | tdasilva: thanks | 15:38 |
| *** d0ugal has quit IRC | 15:38 | |
| *** vinurs has quit IRC | 15:41 | |
| *** d0ugal has joined #openstack-swift | 15:41 | |
| *** sgundur_ has joined #openstack-swift | 15:44 | |
| *** ianstanfield has joined #openstack-swift | 15:46 | |
| *** mvk has quit IRC | 15:47 | |
| *** Jeffrey4l_ has joined #openstack-swift | 15:48 | |
| *** vinsh has joined #openstack-swift | 15:49 | |
| *** ppai has quit IRC | 15:50 | |
| *** ianstanfield has quit IRC | 15:50 | |
| *** chsc has joined #openstack-swift | 15:53 | |
| *** cbartz has joined #openstack-swift | 16:00 | |
| *** oshritf has quit IRC | 16:00 | |
| *** rcernin has quit IRC | 16:07 | |
| acoles | jrichli, mathiasb, mahatic, tdasilva: updated etherpad https://etherpad.openstack.org/p/swift_keymaster | 16:07 |
| jrichli | acoles: thanks for recording results | 16:08 |
| zaitcev | Guys, anyone does some kind of function tests with Hummingbird? I found that it's impossible to tell it to use my test /etc/swift/swift.conf. Seems impossible, but the code just uses configLocations and that's it. | 16:09 |
| zaitcev | I'm going to add another flag | 16:10 |
| *** sgundur_ has quit IRC | 16:15 | |
| openstackgerrit | Takashi Kajinami proposed openstack/swift: Pass logger instances to AccountBroker/ContainerBroker https://review.openstack.org/295875 | 16:20 |
| *** manous has quit IRC | 16:20 | |
| *** tesseract has quit IRC | 16:21 | |
| *** catintheroof has joined #openstack-swift | 16:32 | |
| *** manous has joined #openstack-swift | 16:34 | |
| *** sgundur_ has joined #openstack-swift | 16:46 | |
| *** pcaruana has quit IRC | 16:58 | |
| *** ianstanfield has joined #openstack-swift | 16:59 | |
| *** tqtran has joined #openstack-swift | 17:01 | |
| *** manous has quit IRC | 17:03 | |
| *** tqtran has quit IRC | 17:03 | |
| *** garyj has joined #openstack-swift | 17:03 | |
| *** winggundamth has quit IRC | 17:05 | |
| notmyname | good morning | 17:11 |
| garyj | g'morning | 17:11 |
| *** McMurlock1 has quit IRC | 17:20 | |
| *** LiberalCarrot has joined #openstack-swift | 17:22 | |
| *** LiberalCarrot has left #openstack-swift | 17:22 | |
| *** manous has joined #openstack-swift | 17:25 | |
| *** hseipp has quit IRC | 17:25 | |
| *** sgundur_ has quit IRC | 17:28 | |
| *** sgundur_ has joined #openstack-swift | 17:29 | |
| notmyname | this has some good info on the openstack forum in boston http://superuser.openstack.org/articles/openstack-forum/ | 17:33 |
| *** sams-gleb has quit IRC | 17:33 | |
| *** sams-gleb has joined #openstack-swift | 17:34 | |
| *** sams-gleb has quit IRC | 17:35 | |
| *** sams-gleb has joined #openstack-swift | 17:35 | |
| garyj | notmyname: thanks | 17:40 |
| *** sgundur_ has quit IRC | 17:44 | |
| *** sgundur_ has joined #openstack-swift | 17:44 | |
| *** LiberalSquash has joined #openstack-swift | 17:45 | |
| *** LiberalSquash has left #openstack-swift | 17:45 | |
| *** JimCheung has joined #openstack-swift | 17:55 | |
| *** AnxiousGarlic has joined #openstack-swift | 17:58 | |
| *** AnxiousGarlic has left #openstack-swift | 17:58 | |
| notmyname | acoles: jrichli: et al. thanks for the notes on the etherpad | 17:59 |
| *** rledisez has quit IRC | 17:59 | |
| *** jordanP has quit IRC | 18:04 | |
| *** manous has quit IRC | 18:05 | |
| *** iwaniwan has joined #openstack-swift | 18:12 | |
| iwaniwan | If i use django swift client, can i upload file more than 5 gb with auto segmentation? | 18:14 |
| clayg | acoles: you need to get that line about "review is colloboration not judgement" into the review guidelines -> https://github.com/openstack/swift/blob/master/REVIEW_GUIDELINES.rst | 18:15 |
| notmyname | iwaniwan: if the django client does the segmentation, then sure (in that sense, it's not any different than any other client) | 18:15 |
| clayg | ^ that was some poetical shit man | 18:15 |
| *** manous has joined #openstack-swift | 18:16 | |
| jrichli | wow, did I miss a good dev mailing list thread again? i'll go l ook | 18:19 |
| notmyname | jrichli: https://www.youtube.com/watch?v=kXsIfpnThFA | 18:20 |
| jrichli | notmyname: lol, good use of that one | 18:20 |
| clayg | notmyname: you're so jaded | 18:21 |
| *** cbartz has quit IRC | 18:21 | |
| *** iwaniwan has quit IRC | 18:22 | |
| *** ChubYann has joined #openstack-swift | 18:23 | |
| openstackgerrit | John Dickinson proposed openstack/swift: added a quote https://review.openstack.org/417129 | 18:24 |
| notmyname | clayg: there ^ | 18:24 |
| notmyname | alternatively, "the UK is sleeping, land code referencing acoles!" | 18:24 |
| clayg | heh, we'll we should probably give him a chance to correct my probably flawed retelling of that quote - but yeah - that's the general idea. | 18:25 |
| MooingLemur | is there a published spec for the X-Storage-Pass/X-Storage-User/X-Auth-User/X-Auth-Key on whether those headers should be percent-escaped? swauth seems to unescape them, but ftpcloudfs (via python swiftclient) does not escape input. | 18:26 |
| acoles | clayg: notmyname: sometimes I just watch in silence | 18:26 |
| clayg | ... when he's *sleeping* | 18:26 |
| clayg | acoles: the doug crockford quote seems to just indent? | 18:26 |
| notmyname | MooingLemur: all header values should be url-quoted UTF8 | 18:26 |
| clayg | I like the idea that the author of "javascript the good parts" and alister coles are of similar stature in the eyes of swift reviewers | 18:27 |
| acoles | LOL | 18:28 |
| timburke | notmyname: ...we have func tests that definitely *don't* do that | 18:28 |
| acoles | what does that say about the judgement of swift reviewers? :D | 18:28 |
| notmyname | timburke: well, I'm not sure I actually just said that correctly. mostly, if unicode comes up, I'm confused on what to do where/when | 18:28 |
| timburke | aren't we all :-( | 18:28 |
| clayg | i guess as I understand it Mr. jslint has got himself in a bit of hot water these last few years with SJW's for a few poor taste comments made in jest :\ | 18:29 |
| clayg | acoles would never do that | 18:29 |
| notmyname | maybe I was thinking of header keys | 18:29 |
| acoles | clayg: maybe we could co-author "Swift: the good parts" | 18:29 |
| clayg | lol @ unicode == confused | 18:29 |
| * acoles waits for obvious jokes on that one | 18:30 | |
| *** sgundur_ has quit IRC | 18:30 | |
| clayg | acoles: it'd be A SHORT BOOK LOL! | 18:30 |
| MooingLemur | notmyname: we found that the issue comes up specifically when someone tries to set % inside of a password. They can't set %, but they can set as %25, but then the input to swiftclient's get_auth_1_0 must contain %25 in the key rather than % itself. | 18:30 |
| * acoles didn;t need to wait long | 18:30 | |
| clayg | rofl | 18:30 |
| * clayg wipes tears | 18:30 | |
| clayg | acoles: happy new year btw | 18:30 |
| timburke | acoles: step 1: everything is ascii. step 2: just talk directly to the storage nodes. rip out all that middleware stuff | 18:30 |
| notmyname | clayg: oh hey. you can finally have a stable irc client! https://irssi.org/2017/01/05/irssi-1.0.0-released/ | 18:31 |
| MooingLemur | haha | 18:31 |
| acoles | clayg: you too | 18:31 |
| clayg | i got my swift 2.11 release out (just in time for 2.12) and got distracted by ec reconstructor inefficiencies - think i see the end on that deal and need to push some stuff upstream - so you may get stuck with a bit more of me for a few weeks until the next tire fire | 18:31 |
| timburke | clayg: is that when there'll be another -dev golang ML thread or something? | 18:32 |
| timburke | why does there *have* to be a next tire fire? | 18:33 |
| notmyname | timburke: ...implying it could be the *same* tire fire? ;-) | 18:33 |
| clayg | timburke: those are more like indigestion | 18:34 |
| *** sgundur_ has joined #openstack-swift | 18:35 | |
| *** EmilienM has joined #openstack-swift | 18:37 | |
| EmilienM | hello Swift experts :-) happy new year first | 18:37 |
| notmyname | hello EmilienM | 18:37 |
| EmilienM | I might need some help to understand a failure we have in TripleO CI, in HA environement (3 controllers), with Glance deployed using Swift backend | 18:37 |
| EmilienM | the error happens when we try to upload and list an image in Glance | 18:38 |
| EmilienM | the error in Glance is here: http://logs.openstack.org/74/416974/2/check-tripleo/gate-tripleo-ci-centos-7-ovb-ha/6f2c9a7/logs/overcloud-controller-0/var/log/glance/api.txt.gz#_2017-01-05_17_30_11_243 | 18:38 |
| EmilienM | and you can see the trace in Swift: | 18:38 |
| EmilienM | http://logs.openstack.org/74/416974/2/check-tripleo/gate-tripleo-ci-centos-7-ovb-ha/6f2c9a7/logs/overcloud-controller-0/var/log/swift/swift.txt.gz#_Jan__5_17_30_11 | 18:38 |
| EmilienM | we were wondering if it could be the "node_timeout" value (30s in our setup) that might be too low (our image is very very small though...) | 18:39 |
| EmilienM | note: this problem is inconsistent and doesn't happen all the times (maybe 10%) | 18:39 |
| EmilienM | before jumping on launchpad and file a bug, do you have any thoughts? | 18:39 |
| *** sgundur_ has quit IRC | 18:41 | |
| notmyname | EmilienM: I wonder if there's some sort of deadlock going on. looks like both sides errored on a read | 18:42 |
| EmilienM | if you're wondering since when we have this problem, we use this logstash query: build_name: *tripleo-ci* AND build_status: FAILURE AND message: "Error finding address for" | 18:43 |
| EmilienM | and we think it's something that merged in Swift between Dec 25th and January 3rd | 18:44 |
| notmyname | heh, there were literally zero commits landed between dec 25 and jan 4 | 18:44 |
| *** mvk has joined #openstack-swift | 18:45 | |
| notmyname | https://github.com/openstack/swift/commits/master | 18:45 |
| EmilienM | wait sorry, let me check the exact dates, it might be a bit before | 18:45 |
| *** sgundur_ has joined #openstack-swift | 18:46 | |
| EmilienM | ok it was 23 | 18:46 |
| EmilienM | could it be related then? https://github.com/openstack/swift/commit/cbb99f97b02857dd638b0d61f5e19b584037b25f | 18:47 |
| EmilienM | I don't think so... | 18:47 |
| timburke | what happened on the client side? it looks like they ended the chunked transfer improperly, then glance sent an extra newline or something? | 18:47 |
| *** caiobren_ has joined #openstack-swift | 18:48 | |
| *** garyj has quit IRC | 18:48 | |
| EmilienM | timburke: http://logs.openstack.org/74/416974/2/check-tripleo/gate-tripleo-ci-centos-7-ovb-ha/6f2c9a7/console.html#_2017-01-05_17_30_08_670065 | 18:48 |
| EmilienM | this is what we have for client output | 18:48 |
| *** caiobrentano has quit IRC | 18:49 | |
| *** asettle has quit IRC | 18:49 | |
| notmyname | EmilienM: it's hard to tell without more details about what the client sent on the wire. looks like (from the swift-side error) that chunked transfer encoding was being used and the client didn't follow that protocol. unfortunately, all of that is handled in the wsgi server (eventlet) instead of the wsgi app (swift) so it's really hard to figure out what's going on from the app side. (that is | 18:52 |
| notmyname | , it's eventlet that's doing the chunked encoding parsing and then sending the app a stream of data) | 18:52 |
| EmilienM | notmyname: ok, would it be useful if I provide more debug on what clients runs and the output of what happens? | 18:53 |
| notmyname | EmilienM: ultimately what we all need to look at is the loop in the client that's sending chunks of data on the wire | 18:54 |
| notmyname | I'd hope this is done in glance, but it could be the glance code that calls some sdk that uses some http library, etc | 18:54 |
| acoles | good night | 18:56 |
| *** acoles is now known as acoles_ | 18:57 | |
| EmilienM | notmyname: ok, let me add --debug to the CLIs | 18:57 |
| EmilienM | notmyname: so I can provide more logs on what glanceclient is doing. Sounds good? | 18:57 |
| notmyname | ok | 18:58 |
| timburke | notmyname: well, there are two errors: there's the client <-> glance error (which looks like an unexpected socket close? like, glance expected to be able to read more data but then couldn't?) and there's the glance <-> swift error, which it seems like requests might emit an extra newline as we unwind from the IO error | 18:58 |
| timburke | EmilienM: yeah, i'd expect a --debug around https://github.com/openstack-infra/tripleo-ci/blob/master/scripts/tripleo.sh#L739 to be helpful | 18:58 |
| EmilienM | timburke: you got it! | 18:58 |
| notmyname | timburke: oh, that sounds interesting (if requests is emitting an extra newline) | 18:58 |
| EmilienM | I was about doing that. Cool thanks folks ! | 18:58 |
| timburke | maybe because it's trying to re-use the connection? | 18:59 |
| EmilienM | I'll come back here with more debug | 18:59 |
| timburke | it totally makes sense for glance to try to re-use swiftclient connections under normal operation. but if the client hangs up before the upload completes... i think glance has to just close it and grab a new one on the next request | 19:01 |
| EmilienM | so I proposed https://review.openstack.org/417134 but I hope we'll hit this bug, as like I said, it happens randomly | 19:01 |
| patchbot_ | patch 417134 - openstack-infra/tripleo-ci - DNM: add --debug to glance image upload | 19:01 |
| timburke | (really, requests should as part of the error handling during read) | 19:01 |
| notmyname | EmilienM: ...meaning that you'll "recheck" until it does? | 19:02 |
| EmilienM | notmyname: yes, no choice :( | 19:02 |
| notmyname | any chance of being able to wrap it in that --run-until-failure thing that testr has and do it locally? | 19:03 |
| timburke | i wonder if https://review.openstack.org/#/c/120866/ would help isolate the problem? or just paper over the glance <-> swift half to the point that it becomes really hard to spot :-/ | 19:03 |
| patchbot_ | patch 120866 - glance_store - swift store: Add upload recovery | 19:03 |
| *** manous has quit IRC | 19:04 | |
| *** rcernin has joined #openstack-swift | 19:11 | |
| *** oshritf has joined #openstack-swift | 19:12 | |
| *** manous has joined #openstack-swift | 19:17 | |
| *** nikivi has quit IRC | 19:45 | |
| *** nikivi has joined #openstack-swift | 19:50 | |
| *** chlong has quit IRC | 20:03 | |
| *** _JZ_ has quit IRC | 20:17 | |
| *** ukaynar has joined #openstack-swift | 20:22 | |
| *** sgundur_ has quit IRC | 20:28 | |
| *** _JZ_ has joined #openstack-swift | 20:42 | |
| timburke | oh! i see better now: glance <-> swift is fine, it's the proxy-server <-> object-server connection that goes a bit sideways. proxy logs the 499, closes down the putters, and when the object-server goes to read, it throws the error | 20:45 |
| *** catintheroof has quit IRC | 20:45 | |
| *** xlucas has joined #openstack-swift | 20:48 | |
| *** oshritf has quit IRC | 20:49 | |
| *** sgundur_ has joined #openstack-swift | 20:50 | |
| timburke | torgomatic: you wrote https://github.com/eventlet/eventlet/commit/c3ce3ee -- do you happen to remember whether we ended up *doing* anything with that? we supposedly still work with eventlet 0.17.4 (which predates that); do we actually catch & handle ValueErrors on read somewhere? | 20:50 |
| zaitcev | Oooh crap. Looks like Hummingbird does not work with go 1.7. Lots of panic: inconsistent import: | 20:59 |
| openstackgerrit | Tim Burke proposed openstack/swift: Prevent traceback in object-server on client disconnect https://review.openstack.org/417187 | 21:07 |
| *** xlucas has quit IRC | 21:07 | |
| *** sams-gleb has quit IRC | 21:13 | |
| torgomatic | timburke: not to my knowledge; I think I wanted to handle that in __call__ somewhere, but then never got around to bumping our eventlet dependency past 0.17.4 (or whenever that merged) | 21:17 |
| timburke | torgomatic: the EC path looks great! ^^^ should fix the replicated case | 21:19 |
| *** cdelatte has quit IRC | 21:19 | |
| *** silor has joined #openstack-swift | 21:20 | |
| notmyname | timburke: might be good to explain what you told me for EmilienM, so he's not spending time looking in the wrong place | 21:25 |
| tdasilva | timburke: hi | 21:28 |
| *** silor1 has joined #openstack-swift | 21:29 | |
| *** silor has quit IRC | 21:29 | |
| *** silor1 is now known as silor | 21:29 | |
| tdasilva | timburke: going through your comments on symlink and got puzzled by your comment on SLO. It seems like you tested a manifest that has a symlink as a segment | 21:29 |
| timburke | EmilienM: basically, the swift log was a red herring; the proxy handled everything just fine. the object server didn't handle the proxy's disconnect gracefully, but that didn't impact glance at all. the issue is still the IOError in glance's logs, and that seems to be caused by an unexpected client disconnect | 21:29 |
| timburke | tdasilva: yup! | 21:29 |
| tdasilva | timburke: but doesnt slo have a requirement that segments must have a least 1 byte in size? | 21:29 |
| tdasilva | symlinks must be 0 | 21:29 |
| EmilienM | timburke: ok. What could cause this disconnection? network infra? or something in the client? | 21:30 |
| timburke | tdasilva: i created an SLO with normal segments, then copied the segment elsewhere and replaced it with a symlink to the new location. downloading the SLO would still work just fine (SLO is left of symlink, so the segment gets dereferenced just fine); but COPY or POST (under post-as-copy) would fail | 21:32 |
| tdasilva | tricksy | 21:33 |
| timburke | EmilienM: yes? not entirely sure, really. could be code, could just be transient network troubles | 21:34 |
| EmilienM | mhh | 21:35 |
| *** Jeffrey4l_ has quit IRC | 21:35 | |
| timburke | tdasilva: yes :-) but as i recall from torgomatic's original planning of the thing, it should all totally work | 21:35 |
| timburke | when i related it to notmyname, his reaction was "we haven't gotten rid of post-as-copy yet?" :P | 21:36 |
| tdasilva | timburke: yes, i agree, just trying to figure out how to make it totally work now ;) | 21:36 |
| notmyname | :-) | 21:36 |
| tdasilva | and yeah, my first question was similar | 21:36 |
| tdasilva | when can we get rid of post-as-copy? | 21:36 |
| clayg | tdasilva: do it right now! | 21:37 |
| tdasilva | clayg: it sure would make our lives simpler | 21:38 |
| clayg | we did a release with support for post_as_copy = false that makes sense - this next os release is the time to make post_as_copy = false the default and emit a deprecation warning if you have the wrong value set! | 21:39 |
| timburke | tdasilva: we'd still have to worry about COPY, though | 21:39 |
| notmyname | yep. clayg took the words right out of my mouth | 21:39 |
| notmyname | and furthermore, when we did the original post_as_copy=true release, we said... | 21:40 |
| notmyname | While the default still is | 21:40 |
| notmyname | `object_post_as_copy` as True, the plan is to change the default | 21:40 |
| notmyname | to False and then deprecate post-as-copy functionality in later | 21:40 |
| notmyname | releases. | 21:40 |
| tdasilva | yeah, i think that was me ;) | 21:40 |
| tdasilva | hehe | 21:40 |
| notmyname | that was waaay back in 2.7.0 (Mar 25, 2016) | 21:40 |
| tdasilva | but would we ever remove it all together? | 21:40 |
| notmyname | tdasilva: yes! delete it! from orbit! with fire! | 21:41 |
| clayg | if it's been ~6mo since people have been running a version of swift that calls them basically nazi's for using the wrong setting and no one complained at us I think we can safely assume it's ok to rip it out without making anyone too sad | 21:41 |
| clayg | it's not like it has a client facing behavior chagne where we can't really know what they're doing | 21:42 |
| clayg | if you love post-as-copy so much - DON'T UPGRADE | 21:42 |
| notmyname | we need a release where we change the default and warn if it's set to true | 21:42 |
| tdasilva | does hummingbird have support for post yet? | 21:42 |
| clayg | no | 21:42 |
| clayg | tdasilva: you read the latest interview with taylor rhodes from RAX? | 21:42 |
| tdasilva | notmyname: swift release or OS release? | 21:42 |
| clayg | tdasilva: which ever, it's wall time for majority of swift operators | 21:43 |
| notmyname | tdasilva: OS release. so we need to do it asap in swift and include it in the ...what are we on? ocata? release | 21:43 |
| tdasilva | notmyname: right | 21:44 |
| tdasilva | clayg: did not, link? | 21:44 |
| clayg | potatoe | 21:44 |
| notmyname | then we drop it after the ocata release, asap, and we're good to go | 21:44 |
| clayg | robert is your fathers brother | 21:44 |
| notmyname | claybot got stuck answering riddles | 21:45 |
| *** Jeffrey4l_ has joined #openstack-swift | 21:48 | |
| *** sgundur_ has left #openstack-swift | 21:49 | |
| jrichli | ha! | 21:52 |
| openstackgerrit | Tim Burke proposed openstack/swift: Emit deprecation warnings when running POST-as-COPY https://review.openstack.org/417209 | 22:00 |
| timburke | notmyname: or do you want me to go ahead and change the default, too? | 22:00 |
| notmyname | timburke: yes, please. there will be a handful of docs changes needed too | 22:01 |
| notmyname | and thanks :-) | 22:01 |
| timburke | hmmm... we should consider doing something similar for the allow_versions setting in container-server.conf | 22:04 |
| tdasilva | timburke: you are on fire! | 22:06 |
| tdasilva | don't stop | 22:06 |
| *** darrenc is now known as darrenc_afk | 22:07 | |
| openstackgerrit | Pete Zaitcev proposed openstack/swift: Add the -s flag to specify hummingbird.conf https://review.openstack.org/417216 | 22:15 |
| *** silor has quit IRC | 22:25 | |
| *** darrenc_afk is now known as darrenc | 22:25 | |
| openstackgerrit | Merged openstack/swift: added a quote https://review.openstack.org/417129 | 22:30 |
| pdardeau | what's the backstory on the acoles quote? | 22:34 |
| notmyname | pdardeau: mostly it's what clayg said: acoles is really smart | 22:35 |
| notmyname | it's something he said in a talk and in conversation either in barcelona or before, when we were all talking about contributions, the community, reviews, etc. it stuck out to me (and apparently also to clayg) | 22:36 |
| pdardeau | oic, thx for elaborating. i knew the really smart part already | 22:36 |
| notmyname | this is an interesting discussion https://github.com/golang/go/issues/18517 | 22:37 |
| jrichli | pdardeau: there is a link in a comment to the exact place in the the talk if you visit the review | 22:43 |
| openstackgerrit | Tim Burke proposed openstack/swift: Default object_post_as_copy to False https://review.openstack.org/417209 | 22:43 |
| openstackgerrit | OpenStack Proposal Bot proposed openstack/swift: Updated from global requirements https://review.openstack.org/88736 | 22:44 |
| pdardeau | jrichli: nice! thx for mentioning. i'll watch :) | 22:44 |
| timburke | hmm... should we just drop object_post_as_copy from the manpages entirely? i'm comfortable doing that for the saio example conf, but less so with the manpages. at the same time, the full explanation seems longer than what should be in a manpage... | 22:54 |
| notmyname | timburke: no, it's still a thing, even if there's a warning. I wouldn't remove docs on how it works until it actually isn't a thing any more | 22:56 |
| notmyname | (IMO) | 22:56 |
| *** si1ver has joined #openstack-swift | 22:57 | |
| timburke | notmyname: the manpage currently only describes what happens when you set it to false :-/ since that'll be the default... | 22:58 |
| timburke | and it seems silly to start adding *more* documentation about how post-as-copy works when we don't believe there's a reason to enable it | 22:58 |
| notmyname | timburke: ah | 23:00 |
| notmyname | hmm... | 23:00 |
| timburke | similar issue in the deployment guide... | 23:00 |
| notmyname | FYI a feature idea someone at swiftstack on the customer-facing side would like to see https://blueprints.launchpad.net/swift/+spec/human-readable-tempurl-timestamp | 23:03 |
| *** manous has quit IRC | 23:15 | |
| *** _JZ_ has quit IRC | 23:18 | |
| *** kei_yama has joined #openstack-swift | 23:22 | |
| *** manous has joined #openstack-swift | 23:29 | |
| *** geaaru has quit IRC | 23:29 | |
| *** _JZ__ has quit IRC | 23:31 | |
| clayg | timburke: anywhere you find the docs lacking just say literally "Deprecated" - not "what's the difference" or "this is the behavior if a or b" just "don't use this option" | 23:34 |
| clayg | timburke: if there's existing doco - just make sure it reflects the new/correct default | 23:34 |
| clayg | if someone encounters the deprecated option but still thinks they need to change it - they know where to find us | 23:35 |
| clayg | the bigger issue is getting people currently running the default to not freak out when all of their POSTs are faster and more consistent after they upgrade! | 23:35 |
| *** chsc has quit IRC | 23:38 | |
| *** JimCheung has quit IRC | 23:41 | |
| *** JimCheung has joined #openstack-swift | 23:41 | |
| *** JimCheung has quit IRC | 23:45 | |
| *** klamath has quit IRC | 23:45 | |
| *** mvk has quit IRC | 23:57 | |
| *** manous has quit IRC | 23:59 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!