00:54:25  * johngrastyjoined
01:20:00  * ed209quit (Remote host closed the connection)
01:20:07  * ed209joined
03:21:14  * ryaengjoined
04:13:45  * chatterjoined
04:13:51  <chatter>hey guys
04:14:05  * chatterquit (Client Quit)
04:47:19  <rmustacc>dillona: So if you look at that section of the docs, there are two ways to upload a file. One with an explicit content-length in advance and the other with chunked encoding.
04:48:01  <rmustacc>dillona: If you're using chunked encoding, then you need to esitimate the size for us so we have a sense of how large it is so we can properly place it. Hence why we default to 5 GiB if the header you mentioned isn't specified.
05:08:23  * [0xAA]quit (Ping timeout: 245 seconds)
05:12:58  * elijahZ24quit (Quit: Leaving.)
05:13:05  * elijahZ241joined
08:58:08  * bsmithx10joined
09:40:44  * elijahZ241quit (Quit: Leaving.)
09:40:49  * elijahZ24joined
10:20:00  * ed209quit (Remote host closed the connection)
10:20:07  * ed209joined
12:18:36  * elijahZ24quit (Quit: Leaving.)
12:18:42  * elijahZ24joined
14:06:18  * elijahZ24quit (Quit: Leaving.)
14:06:24  * elijahZ241joined
15:08:58  <dillona>rmustacc: Oh okay. So does that mean that if you want to upload a file > 5 GiB, the header MUST be specified?
15:23:04  <rmustacc>dillona: If you are _streaming_ the file with an unspecified content-length.
15:23:14  <dillona>Right right
15:23:52  <dillona>I think I get it
15:23:52  <dillona>Thanks
16:59:43  <bsmithx10>I'm doing a fresh deployment of manta, I see that the manta services have no owner or package
16:59:49  <bsmithx10>thats normal?
17:05:43  <jayschmidt>bsmithx10: not sure I'm following the question here - what are you seeing?
17:05:51  <jayschmidt>(also, it may just be me being dense today)
17:06:58  <bsmithx10>https://s28.postimg.org/u6wphj9zh/Screenshot_at_2016_12_25_12_06_03.png
17:07:26  <bsmithx10>lol, jayschmidt at least your density is confined into 1 day :P
17:07:42  <bsmithx10>but, you probably aren't being dense.
17:07:44  <jayschmidt>that's just me being optomistic, mate...
17:07:53  <jayschmidt>so, take a look from the api's directly
17:07:58  <jayschmidt>sdc-vmapi /vms for one of them.
17:08:08  <jayschmidt>I bet they do have packages and owners, but it's just adminui being cranky
17:08:43  <jayschmidt>If so, Jay can see if he can reproduce and file a JIRA to get it sorted if it's indeed broken.
17:10:06  <bsmithx10>it created a user called poseidon?
17:10:49  <bsmithx10>this is my first time bringing up manta, I am trying to deploy all the services to the HN
17:10:53  <bsmithx10>except for storage / marlin
17:11:12  <bsmithx10>I have 2 servers with some storage that i am using, (This is a lab)
17:11:15  <jayschmidt>ok
17:11:33  <jayschmidt>So, you know how the engineers like to name things after aquatic creatures?
17:11:43  <jayschmidt>Poseidon, being king of the sea....is also king of manta so to speak.
17:11:50  <bsmithx10>hahh
17:11:52  <jayschmidt>So, that is your superuser account.
17:11:54  <bsmithx10>no i get that :P
17:12:07  <bsmithx10>Just that that magically apeared :P
17:12:20  <bsmithx10>I guesss... being a god... he can just come and go as wanted!
17:12:28  <jayschmidt>it's generated by the manta-init script.
17:12:38  <jayschmidt>whatever email you give it, that's the email that poseidon gets.
17:12:45  <bsmithx10>gotcha, so I used manta-init lab I believe
17:12:49  <jayschmidt>ok
17:13:02  <jayschmidt>In the manta deployment zone.
17:13:03  <bsmithx10>does that have an hard coded deps for joyents lab?
17:13:07  <jayschmidt>not really.
17:13:27  <jayschmidt>it's just to distinguish from coal and from prod, AFAIK
17:13:51  <jayschmidt>The thing you'll want to do to make your life easy is to read through the scripts in the manta deployment zone under /opt/smartdc/manta-deployment
17:13:59  <jayschmidt>(I think that's the dir - or somethign similar)
17:14:08  <bsmithx10>yeah, i read through the deploy-dev
17:14:12  <jayschmidt>ok, brilliant.
17:14:21  <bsmithx10>made a genconfig
17:14:23  <bsmithx10>edited it and deployed
17:14:39  <jayschmidt>did you deploy the nameservers first and let them get settled?
17:15:00  <bsmithx10>I gave thema little
17:15:18  <jayschmidt>That's a point where things can come off the rails.
17:15:29  <bsmithx10>anyway to test if its working?
17:15:33  <jayschmidt>since they have the zookeepers that keep things happy, as well as the binders for internal service discovery.
17:15:44  <bsmithx10>i only have 1 instance
17:16:04  <jayschmidt>so just 1 manatee or just 1 manatee cluster (3 indiv manatees)?
17:16:17  <bsmithx10>i kinda am running wih a coal setup, but with the storage / marlins running on nodes
17:16:27  <jayschmidt>ok
17:16:35  <bsmithx10>but, i can add HA right?
17:16:39  <jayschmidt>yeah
17:16:46  <bsmithx10>I didnt know how big of a hit allthose instances were going to take
17:16:54  <jayschmidt>You've run manta-shardadm?
17:16:54  <bsmithx10>and if I had the dram on the HN to handleit
17:16:58  <jayschmidt>To set your shards?
17:17:25  <bsmithx10>I have 1 1.moray.east.tritonhost.com
17:17:45  <jayschmidt>ok, so you told shardadm to use that for jobs, index, and storage.
17:17:46  <bsmithx10>i pointed a mls with a user at the loadbalancer, but got conn refused
17:18:02  <jayschmidt>and you've generated the hash ring?
17:18:06  <bsmithx10>yeah, i hacked that deploy-lab script
17:18:09  <jayschmidt>ok
17:18:14  <bsmithx10>and ran that
17:18:21  <jayschmidt>So, let's go into the manta ops zone
17:18:23  <jayschmidt>manta-login ops
17:18:26  <bsmithx10>ahhhh
17:18:38  <jayschmidt>that is setup to run everything as poseidon
17:18:48  <jayschmidt>also, it runs your cron jobs for GC, metering, et.
17:18:54  <bsmithx10>i have 2 op zones
17:19:01  <bsmithx10>one on the hn? and another one?
17:19:11  <jayschmidt>ummmm, you should only have one. It's like highlander.
17:20:03  <bsmithx10>https://paste.ec/paste/hQGWVhQD#ThRgvah91-zD8HR6MN54wqLsUP1/e+DVTYKdyH2xmXZ
17:20:38  <bsmithx10>https://paste.ec/paste/lSW2E8lB#a2mqoUM76ZJ0H6rknEbNOu4f6wtJsCtjI1UdzAeiQG-
17:21:13  <jayschmidt>well, that is werid
17:22:25  <jayschmidt>So, take a look at manta-adm show w/o any options (I think - I dont' have a handy manta to validate the command) - find the uuid of the vm for the ops zone and make sure that matches up to one of the two you are seeing in the output above.
17:22:27  <bsmithx10>https://paste.ec/paste/sKqd0Zi+#36DyL1WTia6X7OsTv35AxX2P4wrUrxNcAil1-7Hmd2G
17:22:58  <jayschmidt>ok, so 91835be2-2f38-4ba7-b42d-9d5525feed16 is the *real* ops zone
17:23:10  <jayschmidt>the other one....can you get the uuid it thinks it is?
17:23:16  <jayschmidt>(manta, thinks, that is)
17:23:31  <bsmithx10>Where is that defined?
17:23:33  <bsmithx10>im new
17:23:53  <jayschmidt>should be in manta-adm show output
17:24:13  <bsmithx10>when it tries to login to
17:24:15  <bsmithx10>zlogin: zone '788f6552-4062-4e6e-8a48-145f76ea536a' unknown
17:24:51  <bsmithx10>ok i see whats happening
17:24:56  <bsmithx10>I deployed to a CN
17:25:10  <bsmithx10>and then vmadm list |grep | awk |xargs vmadm detroyedall the zones
17:25:21  <bsmithx10>I never cleaned up the metadata
17:25:31  <bsmithx10>so it has orphaned services "_"
17:25:42  <jayschmidt>yeah, that's - to use a technical term - "not good"
17:25:44  <jayschmidt>:)
17:25:45  <bsmithx10>anyway to clean that up, or start over clean
17:25:53  <jayschmidt>there is a manta-factoryreset
17:25:59  <jayschmidt>you run from the manta deployment zone.
17:26:01  <bsmithx10>k
17:26:06  <jayschmidt>I would do that, which should wipe the slate clean.
17:26:13  <jayschmidt>Then you should be able to redeploy.
17:26:25  <bsmithx10>Error: Attempting to factory reset in a production stage. Failing...
17:26:27  <bsmithx10>wtf
17:26:27  <jayschmidt>As long as you follow along with the coal or lab script you'll be good
17:26:30  <jayschmidt>ah
17:26:34  <bsmithx10>i didnt do anything with production!
17:26:38  <bsmithx10>lol
17:26:51  <jayschmidt>Did you at any time do a manta-init w/ the type set to production?
17:27:18  <bsmithx10>in my history
17:27:32  <bsmithx10>looks like when I was hacking away ( not knowing the rules yet ) :p
17:27:38  <bsmithx10>i init'd a production one
17:27:44  <jayschmidt>yeah
17:27:47  <bsmithx10>i should takea gander at that script
17:27:50  <jayschmidt>that'll do it
17:27:52  <jayschmidt>well, it's node
17:27:55  <jayschmidt>but...the trick is
17:28:04  <jayschmidt> function denyProduction(cb) {
17:28:04  <jayschmidt> var log = self.log;
17:28:04  <jayschmidt> log.info('checking stage');
17:28:04  <jayschmidt> var app = self.application;
17:28:05  <jayschmidt> if (app.metadata.SIZE === 'production') {
17:28:06  <jayschmidt> var m = 'Attempting to factory reset in a ' +
17:28:06  <jayschmidt> 'production stage. Failing...';
17:28:07  <jayschmidt> log.fatal(m);
17:28:07  <jayschmidt> return (cb(new Error(m)));
17:28:08  <jayschmidt> }
17:28:08  <jayschmidt> return (cb(null));
17:28:09  <jayschmidt> },
17:28:11  <jayschmidt>just find where that metadata is set
17:28:12  <jayschmidt>and change it
17:28:20  <jayschmidt>most likely in the manta application somewhere
17:28:30  <jayschmidt>so, sdc-sapi /applications
17:28:49  <bsmithx10>the mdata there Id think
17:29:06  <bsmithx10>size
17:29:17  <jayschmidt># sdc-sapi /applications?name=manta | grep SIZE
17:29:17  <jayschmidt> "SIZE": "production",
17:29:48  <jayschmidt> # sdc-sapi /applications?name=manta | json -Ha metadata.SIZE
17:29:48  <jayschmidt>production
17:30:01  <jayschmidt>so, yeah, just update that metadata, factory reset, and manta merrily along
17:30:14  <bsmithx10>do you knw the post syntax offhand
17:30:27  <bsmithx10>sdc-sapi /applications/uuid size=lab?
17:30:41  <bsmithx10>sdc-sapi /applications/uuid { "size": "lab" } -X POST ?
17:30:45  <jayschmidt>ummm
17:30:49  <jayschmidt>1 sec, let me see
17:31:35  <jayschmidt>ok, it's a put
17:31:37  <jayschmidt>so something like
17:31:38  <jayschmidt>PUT /applications/b0d2f944-7fa3-11e2-a53c-3f3c7a8e7341 -d '{
17:31:38  <jayschmidt> "action": "update",
17:31:38  <jayschmidt> "metadata" {
17:31:38  <jayschmidt> "domain": "lab.joyent.dev"
17:31:38  <jayschmidt> }
17:31:38  <jayschmidt>}'
17:32:07  <jayschmidt>So, you should be able to hack about with that example and use it.
17:32:20  <jayschmidt>It may be easier with sapiadm, but I don't have that syntax handy
17:34:17  <bsmithx10>:)
17:34:24  <bsmithx10>factory-reset going
17:34:44  <jayschmidt>brilliant.
17:34:50  <bsmithx10>but if those zones arent there
17:34:56  <bsmithx10>will it bomb lol
17:34:58  <jayschmidt>it should still wipe the bits from sapi
17:35:01  <bsmithx10>nice
17:35:09  <jayschmidt>If not...then you need to wipe the bits from sap
17:35:12  <jayschmidt>er, sapi
17:35:19  <bsmithx10>Error: The resource in question has "child" elements or is immutable and cannot be destroyed
17:35:20  <jayschmidt>but since it kills the whole application I think it will.
17:35:25  <jayschmidt>nice.
17:35:29  <jayschmidt>so....
17:35:31  <bsmithx10>yeah, ive gotten familiar with sapi
17:35:37  <bsmithx10>by blowing it up
17:35:39  <bsmithx10>:)
17:35:44  <jayschmidt>We've all been there.
17:36:02  <jayschmidt>So, basically, you need to go through and find instances tied to manta in sapi, remove them.
17:36:04  <jayschmidt>Then services
17:36:20  <bsmithx10>the instances are under the manta umbrella?
17:36:21  <jayschmidt>Actually, getting rid of the instances should allow the reset to run
17:36:26  <jayschmidt>they should all be.
17:36:35  <jayschmidt>manta is the overarching application there
17:36:38  <jayschmidt>as opposed to sdc
17:36:52  <bsmithx10>sdc-sapi /applications?name=manta is clean
17:37:08  <bsmithx10>what is sapi based on in node?
17:37:21  <bsmithx10>I was checking out workflow, and vasync yesterday
17:37:38  <bsmithx10>because the company i work for decided to use stackstorm
17:37:43  <bsmithx10>whcih is all python
17:38:04  <bsmithx10>and uses mistral, and I was like.... I know this company who has a decent workflow engine :P
17:38:56  <jayschmidt>You'd have to ask the smart engineers that question - I just kinda kick things to make it work.... ;)
17:39:20  <bsmithx10>As you can see my approach in my lab is
17:39:26  <bsmithx10>HIT THE SODA MACHINE HARD
17:39:50  <jayschmidt>that works too... ;)
17:40:16  <bsmithx10>does the init only create the sapi services?
17:40:32  <bsmithx10>i see now deps on a file or anything
17:40:36  <jayschmidt>I believe so - it creates services and downloads images.
17:40:39  <bsmithx10>just "lab" and "email"
17:40:54  <bsmithx10>this process is starting to make sense now
17:41:17  <jayschmidt>It does after a bit.
17:41:32  <jayschmidt>It's really quite brilliant and elegant.
17:41:42  <bsmithx10>most sdc stuff is
17:41:49  <bsmithx10>except for the zpool sizing of a CN
17:41:58  <bsmithx10>./shoots self.
17:42:20  <jayschmidt>pull requests are always welcome.. :)
17:42:40  <bsmithx10>hahaha, I was talking to someone about how we could introduce that
17:42:48  <bsmithx10>we werethinking in the adminui "name" part
17:43:18  <jayschmidt>honestly, the bulk of server setups we do for our gear and for on-prem gear is done via the CLI
17:43:21  <jayschmidt>(via sdc-server setup)
17:44:40  <bsmithx10>yea, but that depends on the
17:45:03  <bsmithx10>script logic in whats that file name?
17:45:11  <bsmithx10>been awhile since I looked
17:45:18  <bsmithx10>i just usually leave only 4 drives in
17:45:21  <bsmithx10>let it makea raidz
17:45:25  <bsmithx10>add the other 4
17:45:30  <bsmithx10>zpool add
17:46:00  <jayschmidt>disklayout is the bit that determines layouts.
17:46:16  <bsmithx10>Pain threshold, hasnt reached the level to invoke motivation
17:46:58  <bsmithx10>whatsa a vnode?
17:47:08  <bsmithx10>inode except in relation to objects ?
17:47:45  <jayschmidt>In terms of manta or zfs?
17:48:37  <bsmithx10>i guess manta
17:49:32  <bsmithx10>ahh k its VFS stuffs
17:50:37  <bsmithx10>its deploying, shower time!
17:51:09  <jayschmidt>yeah
17:51:11  <jayschmidt>enjoy
17:51:12  <jayschmidt>:)
17:51:42  <bsmithx10>well before I take off
17:51:48  <bsmithx10>is there a link in the operators guide
17:51:51  <bsmithx10>or a way to test this
17:51:58  <bsmithx10>I really have no idea how to add a user or what to do
17:52:00  <jayschmidt>ok
17:52:03  <jayschmidt>so, once it's up
17:52:06  <bsmithx10>my plan was to replace the cert
17:52:09  <jayschmidt>manta-login ops
17:52:10  <bsmithx10>and mls the LB
17:52:14  <jayschmidt>mls ~~/stor
17:52:15  <bsmithx10>loloo, obviously that doesntwork
17:52:18  <jayschmidt>mjob whatever
17:52:23  <jayschmidt>manta uses manta
17:52:31  <jayschmidt>for housekeeping and the like
17:52:39  <bsmithx10>under the posiedon account?
17:52:40  <jayschmidt>So, if you can do things from teh manta ops zone, you've installed it right.
17:52:48  <jayschmidt>yes, it all gets automatically setup for you.
17:52:52  <bsmithx10>nice
17:52:55  <jayschmidt>just get into the ops zone and go from there.
17:52:59  <jayschmidt>Beyond that, it's just normal manta stuff
17:53:18  <bsmithx10>so it inherits all of my users
17:53:33  <bsmithx10>and manta.east.tritonhost.com = the LB?
17:53:41  <jayschmidt>yup
17:53:44  <bsmithx10>the vhost for ha proxy i guess
17:53:58  <bsmithx10>or is that all tcp
17:54:18  <jayschmidt>not 100% sure there
17:54:26  <bsmithx10>ill take a dive
17:54:33  <bsmithx10>Thanks Jay! Im excited,
17:54:42  <bsmithx10>I've got a manta install in my cloudzet
17:54:49  <jayschmidt>:)
17:54:53  <bsmithx10>marlin arejust task runnersright?
17:55:11  <jayschmidt>correct; the marlins run the job. But don't forget that manta uses manta to manage manta
17:55:25  <jayschmidt>if you look at root's crontab in the ops zone
17:55:27  <bsmithx10>so 5 job zonesis ok?
17:55:38  <jayschmidt>should be - you can observe how much they are used.
17:55:38  <bsmithx10>or should i scale that up to 10 per storage cn?
17:55:47  <bsmithx10>ive got 192 gb of ram per cn
17:56:04  <bsmithx10>It shouldn't hurt my Zones too bad while running regular SDC functions right?
17:56:09  <jayschmidt>The standard deployment is ~ 1 per every 2gb
17:56:17  <jayschmidt>The catch is, manta marlins have no cap
17:56:24  <bsmithx10>my lab runs capless
17:56:31  <jayschmidt>ok, you should be fine then.
17:56:34  <bsmithx10>nice
17:56:43  <jayschmidt>Just watch the utilization of the marlins
17:56:47  <bsmithx10>seeya later drop box
17:57:17  <bsmithx10>I needed this too, i wantedto start doing moreautopilot stuff
17:57:35  <bsmithx10>and Noticed Tim and those guys used Manta for a backup target pretty regulary
17:57:48  <jayschmidt>yeah
17:57:58  <jayschmidt>we use it internally for most everything
17:58:14  <bsmithx10>so....
17:58:23  <bsmithx10>i was pondering this
17:58:40  <bsmithx10>would it be possible for you to have a remote delegated dataset?
17:59:03  <bsmithx10>cn1 runs compote agianst a dataset from cn2?
17:59:43  <jayschmidt>Not sure, but the whole design goal of manta was compute co-located with the storage to get rid of the need to muck about like that.
18:00:02  <jayschmidt>when you run a job, manta figures out where your data lives and dispatches accordingly.
18:17:40  <bsmithx10>jayschmidt, it makes sense
18:17:56  <bsmithx10>it really just comes down to, if you need to compute faster
18:18:03  <bsmithx10>you need faster hardware :)
18:18:25  <bsmithx10>jayschmidt, what is mantas fill pattern?
18:30:40  <bsmithx10>jayschmidt, ^C[[email protected] (ops) ~]$ mls ~/stor
18:30:40  <bsmithx10>mls: Error: getaddrinfo ENOTFOUND
19:11:52  * elijahZ241quit (Quit: Leaving.)
19:11:58  * elijahZ24joined
20:20:00  * ed209quit (Remote host closed the connection)
20:20:07  * ed209joined
21:16:54  * elijahZ24quit (Quit: Leaving.)
21:17:00  * elijahZ241joined
21:45:18  * _Tenchi_quit (Quit: Textual IRC Client: www.textualapp.com)
22:20:38  * bahamatjoined
22:55:42  * bahamatquit (Quit: Leaving.)
23:02:49  * yotajoined
23:03:11  * mellocjoined
23:22:14  * mellocquit (Quit: Leaving.)
23:33:06  * [0xAA]joined