| *** rushiagr_away is now known as rushiagr | 02:52 | |
| *** jeromatron has joined #magnetodb | 02:52 | |
| *** rushiagr is now known as rushiagr_away | 03:03 | |
| *** rushiagr_away is now known as rushiagr | 03:05 | |
| *** rushiagr is now known as rushiagr_away | 03:16 | |
| *** jeromatron has quit IRC | 03:21 | |
| *** rushiagr_away is now known as rushiagr | 04:11 | |
| *** ajayaa has joined #magnetodb | 05:18 | |
| rushiagr | I'm having a tough time setting up magnetodb-devstack behind the proxy | 06:03 |
|---|---|---|
| *** ajayaa has quit IRC | 06:11 | |
| *** ajayaa has joined #magnetodb | 06:38 | |
| *** ajayaa has quit IRC | 07:08 | |
| *** ajayaa has joined #magnetodb | 07:21 | |
| *** ajayaa has quit IRC | 07:33 | |
| *** ajayaa has joined #magnetodb | 07:55 | |
| *** idegtiarov has joined #magnetodb | 08:01 | |
| *** idegtiarov has quit IRC | 08:09 | |
| *** idegtiarov has joined #magnetodb | 08:14 | |
| ajayaa | Hi dukhlov, | 08:22 |
| openstackgerrit | Ilya Sviridov proposed a change to stackforge/magnetodb: DevStack installation hanging on ccm stage https://review.openstack.org/119458 | 08:30 |
| dukhlov | Hello | 09:43 |
| dukhlov | Hi, ajayaa | 09:43 |
| ajayaa | dynamodb does not have support for List and dictionary data type it seems. | 09:44 |
| dukhlov | correct | 09:44 |
| ajayaa | Are we providing additional functionality? | 09:44 |
| dukhlov | yes | 09:44 |
| ajayaa | dukhlov, okay. | 09:45 |
| *** isviridov_away is now known as isviridov | 09:45 | |
| isviridov | Hello everybody | 09:45 |
| ajayaa | Hi isviridov | 09:46 |
| isviridov | ajayaa, hello | 09:46 |
| ajayaa | dukhlov, Are we trying to support provisioned throughput as of now? | 09:47 |
| ajayaa | in any way* | 09:48 |
| isviridov | We are not. | 09:48 |
| isviridov | We are not amazon and not all models are applicable for openstack. I mean mdb can be installed on different hardware and reach different performance. | 09:50 |
| isviridov | However, the quota management isin roadmap and it is covers the case mostly. | 09:50 |
| isviridov | BTW we have an opensourced benchmark for mdb. It is not under project ambrela yet, but available for usage. | 09:52 |
| isviridov | Here it is https://github.com/AlexanderChudnovets/magnetodb-test-bench | 09:52 |
| isviridov | rushiagr, what the problem? | 09:56 |
| ajayaa | isviridov, ++ | 09:56 |
| rushiagr | I've pinpointed it to command 'ccm test -v 2.0.8' command | 10:01 |
| rushiagr | even when the proxy environment variables are set, it fails | 10:02 |
| isviridov | rushiagr, how much memory? | 10:02 |
| rushiagr | 4G total | 10:02 |
| isviridov | rushiagr, all other openstack services are also enabled? nova, swift so on? | 10:03 |
| rushiagr | no, only keystone and horizon | 10:04 |
| rushiagr | let me check again. It should've started other services too I think | 10:04 |
| rushiagr | here's the paste http://paste.openstack.org/show/108172/ | 10:05 |
| isviridov | I see it first time. Will you file it as a bug? The usuall problem with memory. Each cassandra node requires 500Mb of memory by default. | 10:17 |
| isviridov | rushiagr, FYI here is a patch where we experementing with memory https://review.openstack.org/#/c/119458/ | 10:18 |
| isviridov | rushiagr, however have you managed to start C* cluster? | 10:19 |
| rushiagr | the ccm itself don't have support for proxy it seems | 10:23 |
| rushiagr | I downloaded cassdandra source, and was trying to run "ant jar" inside it, and it failed | 10:24 |
| rushiagr | this 'ant jar' command I found out by looking into the code of ccm | 10:24 |
| isviridov | rushiagr, not sure about proxy. It is easy to ask in cassandra channel. Have you exported it as env variable? | 10:32 |
| rushiagr | sorry, in a meeting. Will be back in an hour, sorry | 10:32 |
| isviridov | Another problem with ant, i think. You can try to pass JVM proxy setting like that | 10:32 |
| isviridov | export JAVA_OPTIONS='-Dhttp.proxyHost=cache.com -Dhttp.proxyPort=3128' | 10:33 |
| isviridov | rushiagr, sure. See you | 10:33 |
| openstackgerrit | Ilya Sviridov proposed a change to stackforge/magnetodb: Added system memory info output https://review.openstack.org/119726 | 10:50 |
| *** isviridov is now known as isviridov_away | 10:51 | |
| rushiagr | ok, I'm back | 11:10 |
| rushiagr | java_options didn't help | 11:10 |
| rushiagr | isviridov_away: the review you posted is a different error | 12:09 |
| rushiagr | isviridov_away: I'm filing a bug report for my problem, and will include all the details there | 12:09 |
| rushiagr | here it is https://bugs.launchpad.net/magnetodb/+bug/1366788 | 12:15 |
| isviridov_away | rushiagr, I know, I've added more output for shooting https://bugs.launchpad.net/magnetodb/+bug/1364469 | 12:15 |
| rushiagr | isviridov_away: okay | 12:16 |
| isviridov_away | rushiagr, first it is not target usecase for devstack and ccm to work behind proxy server. | 12:18 |
| isviridov_away | rushiagr, but from your output i see that you have problems with DNS | 12:18 |
| rushiagr | isviridov_away: atleast for devstack it is | 12:19 |
| rushiagr | isviridov_away: I have seen bugs related to proxy variables in DevStack | 12:19 |
| rushiagr | and in trove too | 12:19 |
| rushiagr | isviridov_away: DNS I am not sure.. | 12:19 |
| isviridov_away | another idea define -Dhttp.proxyHost as host without 'http://', and make sure that those paramaters are passed to java process. | 12:20 |
| isviridov_away | with checking ps aux| grep java during running ant | 12:20 |
| rushiagr | isviridov_away: okay. Trying.. | 12:20 |
| isviridov_away | About DNS, /home/r/.ccm/repository/2.0.8/build.xml:274: java.net.UnknownHostException: repo2.maven.org looks like it can't resolve the host | 12:22 |
| isviridov_away | could you wget http://repo2.maven.org from you VM? | 12:22 |
| * isviridov_away have to leave right now for 30 mins | 12:23 | |
| rushiagr | isviridov_away: sure, thanks for your help | 12:23 |
| rushiagr | isviridov_away: I can do a wget for sure | 12:23 |
| openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds mechanism for request time measuring https://review.openstack.org/115314 | 12:33 |
| *** isviridov_away is now known as isviridov | 13:03 | |
| openstackgerrit | Ilya Sviridov proposed a change to stackforge/magnetodb: Added system memory info output https://review.openstack.org/119726 | 13:04 |
| isviridov | dukhlov, ikhudoshyn_ please take a look at https://review.openstack.org/#/c/119726/ | 13:07 |
| ikhudoshyn_ | isviridov, will do | 13:07 |
| ajayaa | isviridov, Why dynamodb does not allow attribute definitions which are not present in either primary key and secondary index? | 13:19 |
| ajayaa | https://bugs.launchpad.net/magnetodb/+bug/1268958 | 13:20 |
| dukhlov | ajayaa, because it is not needed for dynamodb | 13:20 |
| dukhlov | but magnetodb is based on cassandra | 13:21 |
| dukhlov | and if you specify another attributes you will use native cassandra support for them | 13:21 |
| ajayaa | dukhlov, okay. | 13:22 |
| dukhlov | and inserting such attributes (predefined) will work faster then another attributes(dynamic) | 13:22 |
| isviridov | ajayaa, good catch, it is not actual anymore. 'Won't fix' | 13:25 |
| ajayaa | dukhlov, Does that mean that in dynamodb I can't have a table with more than six columns? | 13:26 |
| ajayaa | dukhlov, Because it seems that I can't have more than five secondary indices. | 13:27 |
| ajayaa | isviridov, I am confused now. Can I create such a table in dynamodb? | 13:28 |
| ajayaa | Table where all attributes are not present in either primary or secondary index. | 13:29 |
| isviridov | It is up to dynamodb. Have a look at http://docs.aws.amazon.com/amazondynamodb/latest/developerguide/Limits.html | 13:30 |
| isviridov | For magnetodb such limits are very specific to hardware and workload. Benchmarking has to answer for such question. | 13:31 |
| *** bogeyon18 has joined #magnetodb | 13:32 | |
| *** achuprin has joined #magnetodb | 13:34 | |
| isviridov | ajayaa, have I answered? | 13:38 |
| ajayaa | isviridov, I haven't used dynamodb. So I don't know in detail about it. Just read the dynamodb paper. As you invalidated the bug, I can deduce that it is possible to create a table such that all attributes are not necessarily part of indices. | 13:40 |
| *** achuprin has left #magnetodb | 13:42 | |
| *** achuprin has joined #magnetodb | 13:43 | |
| isviridov | ajayaa, not quite right. It is nessesary to mention all indexed attributes, all 'dynamic' attributes shouldn't be desccribed and will be defined during putting data. | 13:43 |
| isviridov | ajayaa, it is possible to get 1 year free tier where dynamodb is included http://aws.amazon.com/free/ | 13:45 |
| isviridov | You need only credit card for this | 13:45 |
| *** achuprin has left #magnetodb | 13:46 | |
| openstackgerrit | Oleksandr Minakov proposed a change to stackforge/magnetodb: Adds count table item in monitoring API https://review.openstack.org/118557 | 13:48 |
| *** SpyRay has joined #magnetodb | 13:49 | |
| isviridov | Hello SpyRay | 13:49 |
| isviridov | SpyRay, welcome back | 13:49 |
| SpyRay | Hi all! | 13:49 |
| *** achuprin has joined #magnetodb | 13:51 | |
| isviridov | SpyRay, rushiagr has trables with installing devstack with mdb behind the proxy | 13:51 |
| isviridov | SpyRay, here it is https://bugs.launchpad.net/magnetodb/+bug/1366788 | 13:51 |
| isviridov | SpyRay, could you take a look? | 13:51 |
| SpyRay | Sure | 13:52 |
| isviridov | gr8! | 13:52 |
| dukhlov | ajayaa, "Does that mean that in dynamodb I can't have a table with more than six columns?" - No | 13:56 |
| dukhlov | you can but you can't specify types of them in table schema at table creation time | 13:57 |
| *** charlesw has joined #magnetodb | 13:58 | |
| isviridov | Hello charlesw | 13:59 |
| charlesw | Hi isviridov | 13:59 |
| isviridov | Hello ominakov. How are you today? | 14:01 |
| ominakov | hello guys | 14:02 |
| ominakov | hello isviridov, i'am fine | 14:02 |
| openstackgerrit | Ilya Sviridov proposed a change to stackforge/magnetodb: (WIP)Adds count table item in monitoring API https://review.openstack.org/118557 | 14:02 |
| isviridov | ominakov, any success with getting table size? | 14:03 |
| isviridov | Do we know how to get it from C*? | 14:03 |
| ominakov | unfortunately no, I can't find this information in system table | 14:04 |
| isviridov | What do guys from cassadnra say? | 14:05 |
| ominakov | i've asked this question in C* irc channel, but not received any response | 14:06 |
| openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Fixes bug with put item predefined attributes https://review.openstack.org/119462 | 14:19 |
| *** ajayaa has quit IRC | 14:29 | |
| *** bogeyon18 has quit IRC | 14:35 | |
| *** isviridov is now known as isviridov_away | 14:40 | |
| charlesw | I don't believe C* maintains a row or key count on table. CQL's select count(*) from has a limit and can take very long time if limit is too high. nodetool cfstats can provide estimate of number of keys but there's no accurate count. | 14:41 |
| openstackgerrit | A change was merged to stackforge/magnetodb: Added system memory info output https://review.openstack.org/119726 | 14:42 |
| openstackgerrit | Alexei Vinogradov proposed a change to stackforge/magnetodb: Heap memory parameters are configurable now https://review.openstack.org/119458 | 14:46 |
| *** SpyRay has quit IRC | 14:49 | |
| *** jeromatron has joined #magnetodb | 15:21 | |
| isviridov_away | jeromatron, hello | 15:23 |
| *** isviridov_away is now known as isviridov | 15:23 | |
| isviridov | jeromatron, we need your help with C* | 15:24 |
| isviridov | :) | 15:24 |
| *** jeromatron has quit IRC | 15:24 | |
| *** jeromatron has joined #magnetodb | 15:24 | |
| jeromatron | okay, what's up? | 15:25 |
| isviridov | You know we are working on building monitoring api in mdb | 15:25 |
| isviridov | The thing is we need to expose size of table in bytes | 15:25 |
| isviridov | Is there any way in cassadnra to get such info? | 15:26 |
| *** jeromatron has quit IRC | 15:27 | |
| * isviridov can't catch this guy | 15:32 | |
| *** bogeyon18 has joined #magnetodb | 15:33 | |
| *** jeromatron has joined #magnetodb | 15:36 | |
| isviridov | jeromatron, I can't catch you :) | 15:37 |
| jeromatron | I'll be in the office shortly. the only way I know is through jmx, getting the column family stats on each node and add those up but that's not 100% accurate. | 15:37 |
| jeromatron | the opscenter guys collect some of that and they expose it through their rest api. I can check with them on that | 15:37 |
| jeromatron | brb | 15:37 |
| *** jeromatron has quit IRC | 15:37 | |
| openstackgerrit | Oleksandr Minakov proposed a change to stackforge/magnetodb: (WIP) Adds count table item in monitoring API https://review.openstack.org/118557 | 15:40 |
| *** isviridov is now known as isviridov_away | 16:13 | |
| *** rushiagr is now known as rushiagr_away | 16:14 | |
| *** bogeyon18 has quit IRC | 16:35 | |
| *** jeromatron has joined #magnetodb | 16:39 | |
| *** jeromatron has quit IRC | 16:46 | |
| *** jeromatron has joined #magnetodb | 16:55 | |
| *** jeromatron has quit IRC | 17:00 | |
| *** jeromatron has joined #magnetodb | 17:02 | |
| *** bogeyon18 has joined #magnetodb | 17:06 | |
| *** rushiagr_away is now known as rushiagr | 17:06 | |
| *** jeromatron has quit IRC | 17:17 | |
| *** jeromatron has joined #magnetodb | 17:17 | |
| *** jeromatron has quit IRC | 17:19 | |
| rushiagr | thanks guys for triaging the bug, and targetting it for j-rc1. Very much needed for our evaluation efforts.. | 17:24 |
| openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds mechanism for request time measuring https://review.openstack.org/115314 | 17:30 |
| *** bogeyon18 has quit IRC | 17:32 | |
| *** bogeyon18 has joined #magnetodb | 17:37 | |
| openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds mechanism for request time measuring https://review.openstack.org/115314 | 17:42 |
| *** jeromatron has joined #magnetodb | 17:44 | |
| openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds mechanism for request time measuring https://review.openstack.org/115314 | 18:08 |
| openstackgerrit | Andrei V. Ostapenko proposed a change to stackforge/magnetodb: Adds mechanism for request time measuring https://review.openstack.org/115314 | 18:19 |
| *** rushiagr is now known as rushiagr_away | 18:32 | |
| *** openstackgerrit has quit IRC | 18:51 | |
| jeromatron | isviridov_away just checking with the opscenter guys now. it sounds like that's what they do. just get the jmx estimate and add it up through the cluster. | 19:03 |
| *** openstackgerrit has joined #magnetodb | 19:08 | |
| *** jeromatron has quit IRC | 19:15 | |
| *** vnaboychenko has joined #magnetodb | 19:59 | |
| *** jeromatron has joined #magnetodb | 20:07 | |
| *** jeromatron has quit IRC | 20:51 | |
| openstackgerrit | Oleksandr Minakov proposed a change to stackforge/magnetodb: (WIP) Adds count table item in monitoring API https://review.openstack.org/118557 | 20:58 |
| *** jeromatron has joined #magnetodb | 21:00 | |
| *** bogeyon18 has quit IRC | 21:28 | |
| *** charlesw has quit IRC | 22:09 | |
| *** bogeyon18 has joined #magnetodb | 22:41 | |
| *** vnaboychenko has quit IRC | 23:37 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!