00:11:31  * eVikingquit (Remote host closed the connection)
00:14:56  * fredkquit (Quit: Leaving.)
00:36:36  <cxe1>so I have two nodes
00:36:49  <cxe1>compute nodes, I want to do marlin calculation on them
00:36:57  <cxe1>but they are not storage nodes
00:37:16  <cxe1>they have a lot of memory and CPU but very little storage
00:38:18  <cxe1>looks like until I provision manta-storage image on them they are not showing up in Marlin-dashboard
00:38:40  <rmustacc>marlin doesn't work that way.
00:38:46  <rmustacc>The design is such that it does compute on the data itself.
00:39:02  <rmustacc>So it cannot run marlin zones on machines without said data, hence why they need a mako.
00:39:24  <cxe1>Ah. so marlin and storage have to go together
00:39:43  <cxe1>what in this case compute nodes are?
00:40:36  <rmustacc>A compute node is generally used as a generic term in SDC that refers to any server that is not the head node.
00:40:44  <rmustacc>There may be something else that you're referring to.
00:41:12  <cxe1>ah. this is where confusion came from.
00:41:26  <cxe1>So all Manta nodes by default are storage nodes
00:42:07  * eVikingjoined
00:42:40  <cxe1>I misinterpreted the following "Storage nodes are compute nodes that are designated to store actual Manta objects. These are the same servers that run users' compute jobs, but we don't call those compute nodes because that would be confusing with the SDC terminology."
00:43:17  <rmustacc>No, that's not correct.
00:43:22  <rmustacc>Not all manta nodes are storage nodes by default.
00:43:45  <cxe1>Ok. All storage nodes participating in marlin are storage nodes
00:43:46  <rmustacc>For example, the nodes that are running the indexing tier, load balancers, databases, etc. are not generally the same as storage nodes.
00:44:11  <rmustacc>Any storage node runs marlin any marlin node has storage.
00:45:38  <cxe1>in this case what is number of storage images on the storage node means. For example config say "storage" "UUID" : number
00:45:57  <cxe1>what is the number means? I understand it is number of zones with UUID image
00:46:06  <cxe1>but why do we need more than one per node?
00:46:20  * eVikingquit (Ping timeout: 246 seconds)
01:20:01  * ed209quit (Remote host closed the connection)
01:20:08  * ed209joined
01:42:42  * eVikingjoined
01:44:34  * bahamatquit (Quit: Leaving.)
01:47:16  * eVikingquit (Ping timeout: 272 seconds)
02:43:16  * eVikingjoined
02:48:00  * eVikingquit (Ping timeout: 276 seconds)
03:08:56  * cxe1quit (Quit: Leaving)
03:16:26  * marsellquit (Ping timeout: 252 seconds)
03:26:51  * bahamatjoined
03:28:01  * bahamatquit (Client Quit)
03:43:50  * eVikingjoined
03:47:59  * eVikingquit (Ping timeout: 246 seconds)
04:44:25  * eVikingjoined
04:49:10  * eVikingquit (Ping timeout: 272 seconds)
05:44:59  * eVikingjoined
05:49:32  * eVikingquit (Ping timeout: 256 seconds)
05:53:19  * klyntonquit (Quit: Connection closed for inactivity)
06:45:31  * eVikingjoined
06:50:19  * eVikingquit (Ping timeout: 256 seconds)
07:46:05  * eVikingjoined
07:50:48  * eVikingquit (Ping timeout: 272 seconds)
08:46:39  * eVikingjoined
08:50:56  * eVikingquit (Ping timeout: 240 seconds)
08:55:47  * bsdguruquit (Quit: bsdguru)
09:17:38  * bsdgurujoined
09:39:57  * pmooneyjoined
09:47:17  * eVikingjoined
09:51:39  * eVikingquit (Ping timeout: 264 seconds)
10:03:34  * pmooneyquit (Ping timeout: 255 seconds)
10:20:00  * ed209quit (Remote host closed the connection)
10:20:07  * ed209joined
10:47:49  * eVikingjoined
10:52:34  * eVikingquit (Ping timeout: 272 seconds)
10:58:16  * bsdguruquit (Quit: bsdguru)
11:17:37  * bsdgurujoined
11:30:21  * marselljoined
11:48:21  * eVikingjoined
11:52:55  * eVikingquit (Ping timeout: 256 seconds)
12:48:54  * eVikingjoined
12:53:24  * eVikingquit (Ping timeout: 256 seconds)
12:54:56  * bsdguruquit (Quit: bsdguru)
13:09:03  * kapil___joined
13:17:28  * bsdgurujoined
13:28:58  * chorrelljoined
13:49:31  * eVikingjoined
13:53:58  * eVikingquit (Ping timeout: 255 seconds)
14:35:59  * chorrellquit (Quit: Textual IRC Client: www.textualapp.com)
14:39:09  * chorrelljoined
14:50:08  * eVikingjoined
14:50:33  * bsdguru_joined
14:50:59  * ryancnelsonjoined
14:53:24  * bsdguruquit (Ping timeout: 276 seconds)
14:54:56  * bsdguru_quit (Ping timeout: 252 seconds)
14:57:17  * cxe1joined
15:04:08  * eVikingquit (Ping timeout: 244 seconds)
15:11:30  * ryancnelsonquit (Quit: Leaving.)
15:30:29  <cxe1>Found the issue with manta-deploy-dev. There is a line where script should stop and wait for the nameserver, but it just keep running screwing up the rest of the install
15:30:56  <rmustacc>Then please write up the details in the form of a bug report.
15:32:24  <cxe1>how would I do that sir? I would love to
15:34:12  <chorrell>https://github.com/joyent/manta/issues
15:38:20  <cxe1>thank you. will do
16:01:37  * eVikingjoined
16:06:08  * eVikingquit (Ping timeout: 265 seconds)
16:07:20  * bsdgurujoined
16:10:39  * happy-dudejoined
16:14:28  * fredkjoined
16:29:44  * dap_joined
16:51:24  * bahamatjoined
17:02:10  * eVikingjoined
17:06:35  * eVikingquit (Ping timeout: 244 seconds)
17:12:23  * chorrellquit (Quit: My Mac has gone to sleep. ZZZzzz…)
18:02:45  * eVikingjoined
18:04:44  * axisyspart
18:07:29  * eVikingquit (Ping timeout: 256 seconds)
18:17:22  * chorrelljoined
19:03:20  * eVikingjoined
19:08:13  * eVikingquit (Ping timeout: 264 seconds)
19:10:17  * chorrellquit (Quit: Textual IRC Client: www.textualapp.com)
19:30:06  <nahamu>if I want to test a Manta installation on CoaL, how big do I need to make the virtual disk that holds the zones pool?
19:30:28  <nahamu>(and do I need to create a second VM or can I do it all on the single headnode VM?)
19:32:06  <jayschmidt>you can do it all on the single headnode vm.
19:32:13  <jayschmidt>that's how I do it.
19:33:54  <nahamu>is there a Manta equivalent to the SDC answers.json that automates setting up Manta on the CoaL VM?
19:34:08  <jayschmidt>1 sec, let me find it for you.
19:35:08  <nahamu>is it just the gen-coal.sh thing?
19:35:26  <jayschmidt>there is one script you can run once you get the HN booted up that just deploys a manta in coal.
19:35:29  <nahamu>https://github.com/joyent/manta/blob/master/docs/manta-ops.md#deploying-manta 3a
19:35:45  <nahamu>that sounds even better if you can find it.
19:36:39  <jayschmidt>hmmm…maybe it's only in coal builds (I'm on my real lab).
19:36:49  <jayschmidt>1 sec. let me find a coal build.
19:37:59  <jayschmidt>ah, ok
19:38:08  <jayschmidt>So, on the usb stick image for your coal.
19:38:11  <jayschmidt>Look under "devtools"
19:38:19  <jayschmidt>Then look at deploy_coal_manta.sh
19:38:31  <nahamu>is that only on the coal usb stick?
19:38:52  <jayschmidt>That will work - the only time I have had issues in the past is with the marlin image download timing out (that's the 13 gb image).
19:38:54  <jayschmidt>yes.
19:38:56  <jayschmidt>Only on coal.
19:39:03  <jayschmidt>It doesn't seem to be on my usbkey for my HN.
19:39:33  <nahamu>I was trying to figure out what was different between the coal image and the regular image but I never looked there.
19:40:01  <nahamu>I didn't see an answers.json file so I just assumed they were identical now. good to know there's a reason to actually unpack the coal tarball after all.
19:40:19  <nahamu>I may be back with more questions later... thanks jayschmidt!
19:40:24  <jayschmidt>no worries, good luck.
19:40:51  <nahamu>oh yeah, and is 25GB for the zones pool enough or should I make it bigger?
19:41:19  <jayschmidt>I'm trying to remember what I did with mine - I haven't deployed Coal/Manta in a while, but I am pretty sure I made mine bigger.
19:49:06  * ryancnelsonjoined
20:03:52  * eVikingjoined
20:06:56  <cxe1>nahamu I can help. I killed 2 weeks perfecting this craft :-)
20:08:13  <cxe1>By default CoaL created 60Gb disk for VM
20:08:20  * eVikingquit (Ping timeout: 246 seconds)
20:11:59  <cxe1>without getting into political discussion is S3 AWS API was even considered for Manta? If I want to migrate existing AWS customers to Joyent I cannot ask them to rewrite their app.
20:14:10  <_Tenchi_>maybe you could write a proxy? :)
20:14:33  <_Tenchi_>maybe one already exists... worth a few minutes to scan joyent's github
20:17:29  <ryancnelson>no, it wasn't considered.
20:18:06  <ryancnelson>Joyent hasn't written one.
20:19:09  <ryancnelson>manta-nfs exists, though. you can mount your manta storage as an nfs filesystem
20:20:01  * ed209quit (Remote host closed the connection)
20:20:08  * ed209joined
20:22:56  <nahamu>cxe1: thanks. I'll give my VM a 60GB disk.
20:25:29  <cxe1><ryancnelson> is hardly REST API substitute. If customer has thousands of line code application it would be very difficult to convince them to rewrite it for Joyent API
20:25:47  <ryancnelson>you said "without getting into political discussion"
20:25:48  <ryancnelson>so.
20:25:56  <cxe1>:-)
20:26:12  <ryancnelson>manta isn't S3
20:27:47  <ryancnelson>a customer shouldn't need to rewrite "thousands of lines of code", they'd likely need to change a handful of lines.
20:27:51  <cxe1>I realize that but seems like with REST API somebody wold already write a gateway for S3. I just didn't want to reinvent the wheel
20:28:57  <ryancnelson>... when performance is bad because of that gateway, people will blame manta. Joyent doing that work would imply "we think this is a good idea", imho
20:29:28  <ryancnelson>plenty of standard object-storage clients already understand both s3 and others, like manta, dreamobjects, box, etc
20:38:08  <cxe1>which client understand both S3 and Manta Joyent?
20:38:19  <cxe1>It might me a quick solution for migration
20:40:21  <ryancnelson>i'd thought libFog was one, at least. I'm not a ruby dev, though.
20:41:14  <ryancnelson>so you're not going to convince me to write an S3-compatible api this afternoon, regardless of its merits\
20:45:59  <dap_>Looks like http://dannysu.com/2013/07/01/introducing-csync/ also speaks both.
20:47:06  * jayschmidtquit (Quit: Leaving.)
21:04:27  * eVikingjoined
21:08:54  * eVikingquit (Ping timeout: 244 seconds)
21:40:23  * bahamatquit (Quit: Leaving.)
21:48:13  * bahamatjoined
21:52:00  * pmooneyjoined
21:57:13  * dap_quit (Quit: Leaving.)
22:01:19  * pmooneyquit (Ping timeout: 255 seconds)
22:05:03  * eVikingjoined
22:05:34  * pmooneyjoined
22:09:21  * eVikingquit (Ping timeout: 244 seconds)
22:10:12  * pmooneyquit (Ping timeout: 276 seconds)
22:24:10  <cxe1>what define number of marlin and storage zones ?
22:24:27  <cxe1>how do I choose how many do I need on the storage node?
23:05:36  * eVikingjoined
23:09:48  * eVikingquit (Ping timeout: 244 seconds)
23:36:28  * happy-dudequit (Quit: Connection closed for inactivity)