*** dave-mccowan has quit IRC | 03:38 | |
*** openstackgerrit has quit IRC | 06:07 | |
*** hoonetorg has quit IRC | 06:39 | |
*** hoonetorg has joined #openstack-sahara | 06:46 | |
*** pcaruana has joined #openstack-sahara | 07:14 | |
*** rcernin has quit IRC | 07:20 | |
*** tosky has joined #openstack-sahara | 07:38 | |
*** hoonetorg has quit IRC | 07:55 | |
*** Bhujay has joined #openstack-sahara | 08:37 | |
Bhujay | tosky , regd image_argeuments NotImplemented errro that I was facing with hdp2.5 , the solution was there https://review.openstack.org/#/q/I88c2a553512a797099a4b2fbd9e9d204475e755b | 08:39 |
---|---|---|
remix_tj | i'm having this error on newton: Creating cluster failed for the following reason(s): RemoteCommandException: Error during command execution: "sudo su - -c "hadoop fs -put -p /usr/java/jdk1.7.0_67-cloudera/jre/lib/tools.jar /user/sahara-hbase-lib" hdfs" Return code: 1 STDERR: put: `/usr/java/jdk1.7.0_67-cloudera/jre/lib/tools.jar': No such file or directory | 09:04 |
remix_tj | this because /usr/java/jdk1.7.0_67-cloudera/jre/lib/tools.jar doesn't exists. Instead /usr/java/jdk1.7.0_67-cloudera/lib/tools.jar exists | 09:04 |
remix_tj | is something the user has failed on input or is a sahara fixed value? | 09:05 |
remix_tj | it appears a wrong value of hbase classpath that sahara is parsing | 09:53 |
tosky | Bhujay: uh, but that commit was already in the last tagged version for queens (8.0.1) | 10:04 |
tosky | remix_tj: is this a centos or a ubuntu image? | 10:05 |
tosky | I guess created with sahara-image-elements? | 10:05 |
remix_tj | tosky: centos image | 10:05 |
remix_tj | but i think is a bit old | 10:05 |
remix_tj | it has been uploaded on 2017-03-28, i'll check a newer one | 10:07 |
tosky | remix_tj: I would try to rebuild a new one | 10:07 |
Bhujay | tosky , it shd be but when i was checking commit history from github tag 8.0.1 could see it . Also in my env the commit was not ther although i am on 8.0.1 may be opnstack ansible was not pointing correctly | 10:08 |
tosky | remix_tj: another idea is that there may have few changes in the jdk package from cloudera which happened afterwards | 10:08 |
Bhujay | could not ** | 10:08 |
remix_tj | the fact is that hbase classpath returns /usr/java/jdk1.7.0_67-cloudera/jre/lib/tools.jar but it doesn't exist, the only tools.jar is /usr/java/jdk1.7.0_67-cloudera/lib/tools.jar | 10:08 |
remix_tj | tosky: the image i've donwloaded is the one distributed by mirantis here http://sahara-files.mirantis.com/images/upstream/newton/sahara-newton-cloudera-5.7.0-centos7.qcow2 | 10:13 |
remix_tj | i think is already ok. But i'll test again without hbase | 10:14 |
tosky | remix_tj: don't rely too much on those | 10:14 |
tosky | I would try to generate it manually | 10:14 |
remix_tj | ok, docs? | 10:14 |
tosky | we need to refactor it, but https://docs.openstack.org/sahara/latest/user/cdh-imagebuilder.html | 10:15 |
tosky | if you use RDO, the sahara-image-elements package is packaged | 10:15 |
remix_tj | yes i use | 10:15 |
tosky | so sahara-image-elements -p cloudera -v 5.7 -i centos7 | 10:19 |
remix_tj | ok, but i have to dig around for newton repos, because i have an ancient cluster | 10:20 |
tosky | or you can use the sahara-image-elements git repository, but you need to checkout the newton-eol tag | 10:23 |
tosky | please remember that you should run sahara-image-elements, and generally everything based on diskimage-elements, on a separate dedicated system | 10:23 |
tosky | because it requires root permissions and it does some magic with loopback images which may go wrong | 10:24 |
tosky | my suggestion is simply run it on a virtual machine | 10:24 |
*** dave-mccowan has joined #openstack-sahara | 10:35 | |
tosky | Bhujay: even thinking about, I really don't know how it could happen; if you still have the openstack-ansible logs from the deployment, maybe they can explain something | 10:35 |
tosky | otherwise... I don't know :) | 10:35 |
remix_tj | tosky: i'm running on the tripleo host, and it is working and placing everything in /tmp | 10:38 |
Bhujay | sure , agreed . I will analyse from the OSA perspective , basicall they point to a particular commit id with every release . So its not a big deal now , and certainly not a sahara bug | 10:40 |
tosky | yeah, we fixed the bug :) | 10:42 |
tosky | remix_tj: uh, if it works... but it's not what I advise to do | 10:42 |
remix_tj | i've seen that specifying 5.7 has selected 5.7.0, what do i need to do for having latest microversion? using the env variable? | 10:43 |
tosky | DIB_CDH_MINOR_VERSION | 10:47 |
remix_tj | ook | 10:48 |
remix_tj | at the moment i test with this version. In case i upgrade | 10:48 |
*** rnm has joined #openstack-sahara | 11:34 | |
*** rnm is now known as rmart04 | 11:36 | |
*** rmart04 has quit IRC | 11:40 | |
*** rnm has joined #openstack-sahara | 11:40 | |
*** rnm has quit IRC | 11:42 | |
*** rnm has joined #openstack-sahara | 11:42 | |
*** rnm is now known as rmart04 | 11:43 | |
remix_tj | tosky: i'm expecting the same error even with a newer image | 13:12 |
remix_tj | because hbase classpath still refers to a non existing jar file | 13:12 |
tosky | uhm | 13:13 |
tosky | I remember I tested it when it was Newton time | 13:13 |
tosky | maybe not hbase though | 13:13 |
remix_tj | let me see if the code has changed through time | 13:13 |
tosky | we changed the image generation process | 13:13 |
remix_tj | but is not an image issue, i think is something related to packages | 13:14 |
tosky | yeah | 13:14 |
tosky | so, hbase was probably not tested | 13:14 |
remix_tj | https://github.com/openstack/sahara/blob/90ca15859e50dc83d80d9411e86a7983ca6914b9/sahara/service/edp/hdfs_helper.py#L32 is the function failing | 13:16 |
remix_tj | tosky: same error, i'll check again | 13:22 |
remix_tj | if not, i'll fix directly the image | 13:28 |
tosky | probably easier | 13:28 |
tosky | I should test it with newer CDH packages | 13:29 |
*** zchkun has joined #openstack-sahara | 13:30 | |
remix_tj | i'd test, but i've newton running | 13:32 |
remix_tj | tosky: found, is a cloudera thing. /usr/lib/hbase/bin/hbase contains CLASSPATH=${CLASSPATH}:$JAVA_HOME/lib/tools.jar, and JAVA_HOME is set to /usr/java/jdk1.7.0_67-cloudera/jre, so thats why the classpath has wrong files | 13:43 |
remix_tj | s/wrong/non existing/ | 13:43 |
*** tellesnobrega has joined #openstack-sahara | 13:44 | |
*** Bhujay has quit IRC | 13:47 | |
*** jeremyfreudberg has joined #openstack-sahara | 13:55 | |
*** zchkun has quit IRC | 14:34 | |
*** jeremyfreudberg has quit IRC | 14:44 | |
*** openstackgerrit has joined #openstack-sahara | 14:58 | |
openstackgerrit | Luigi Toscano proposed openstack/sahara-tests master: Refresh test templates (also for Rocky) https://review.openstack.org/595786 | 14:58 |
*** pcaruana has quit IRC | 15:12 | |
*** rmart04 has quit IRC | 16:41 | |
openstackgerrit | Alex Schultz proposed openstack/puppet-sahara master: Bump version for the start of Stein https://review.openstack.org/598288 | 18:11 |
*** rmart04 has joined #openstack-sahara | 20:47 | |
*** rmart04 has quit IRC | 21:16 | |
*** rcernin has joined #openstack-sahara | 21:49 | |
*** tosky has quit IRC | 22:23 | |
*** tellesnobrega has quit IRC | 22:27 | |
*** rcernin has quit IRC | 23:56 | |
*** rcernin has joined #openstack-sahara | 23:56 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!