*** ukaynar has quit IRC | 00:51 | |
*** ukaynar has joined #openstack-sahara | 00:51 | |
*** ukaynar has quit IRC | 00:56 | |
*** ukaynar_ has joined #openstack-sahara | 03:20 | |
*** rickflare2 has joined #openstack-sahara | 06:11 | |
*** rickflare has quit IRC | 06:15 | |
*** ukaynar_ has quit IRC | 07:20 | |
*** ukaynar has joined #openstack-sahara | 07:21 | |
*** ukaynar has quit IRC | 07:25 | |
*** tesseract has joined #openstack-sahara | 08:16 | |
frippe75 | bad idea having hostname return FQDN ?? Having issues with sahara not being able to spawn the nodes required by the template. Getting "NoValidHost_Remote: No valid host was found" in nova-conductor. I can create instances via "openstack server create" | 08:39 |
---|---|---|
*** ukaynar has joined #openstack-sahara | 09:22 | |
*** ukaynar has quit IRC | 09:26 | |
ltosky[m] | No valid host was found is a generic error in nova and it may mean a lot of problems | 09:26 |
ltosky[m] | Maybe not enough resources for all hosts required | 09:27 |
ltosky[m] | check in the nova-compute logs and see if where the allocation process started, if any node was created and then at some point something failed | 09:28 |
ltosky[m] | I don't think it's related to DNS though | 09:28 |
ltosky[m] | yeah, nova related, not sahara related; at that point, sahara is just using heat to create few instances | 09:31 |
ltosky[m] | there are many help requests about people hitting this (unfortunately) generic error, for example | 09:32 |
ltosky[m] | https://ask.openstack.org/en/question/107755/no-valid-host-was-found-there-are-not-enough-hosts-available/ | 09:32 |
ltosky[m] | frippe75 ^^ | 09:32 |
frippe75 | ltosky[m], thanks for replying! But as I wrote, I can spin up a heat stack that create instances (2 of the using m1.medium). Removing them and recreating my sahara-spark.2.1.0 cluster using 1 master and 1 slave (both m1.medium) gets me the "No valid hosts" | 09:34 |
frippe75 | Cannot see any filtering events telling me why it gets filtered out | 09:35 |
frippe75 | difference between my "heat stack" and the stack from sahara is that the sahara one has cinder volumes. Tried creating the same amount using "openstack volume create" is successful. | 09:36 |
ltosky[m] | and with volumes created by the heat template? | 09:36 |
ltosky[m] | raise the debug also for heat and not only for nova | 09:37 |
frippe75 | Ok I will try that! | 09:54 |
*** ukaynar has joined #openstack-sahara | 11:22 | |
*** ukaynar has quit IRC | 11:27 | |
*** ukaynar has joined #openstack-sahara | 13:23 | |
*** ukaynar has quit IRC | 13:27 | |
*** ukaynar has joined #openstack-sahara | 15:24 | |
*** ukaynar has quit IRC | 15:29 | |
frippe75 | ltosky[m], increasing to debug did just clutter the troubleshooting for me. Went back to looking at sahara config guide for pike and what was actually configured by packstack (pike). under [keystone_authtoken] the guide states I should have auth_uri and identity_uri configured. But I have auth_uri and auth_url set .... problem? | 16:17 |
frippe75 | This is the error I think is most important during the cluster deploy from sahara-engine.log | 16:20 |
frippe75 | CREATE failed: ResourceInError: resources.spark-master.resources[0].resources.spark-master.resources[0].resources.volume: Went to status error due to "Unknown" | 16:20 |
*** ukaynar has joined #openstack-sahara | 16:22 | |
*** tosky has joined #openstack-sahara | 16:25 | |
tosky | the double thing on keystone_authtoken is not a problem there, you would see a different error elsewhere | 16:26 |
tosky | from heat you should see the template generated | 16:27 |
tosky | did you try to create a cluster without attached volumes? | 16:27 |
tosky | frippe75: ^^ | 16:27 |
frippe75 | will try. This worked right out of the box on newton using packstack and my cluster templates. What's changed is that I build my images using sahara-image-elements and moved to newer version of the spark plugin. But I will try without the volumes if thats possible. | 16:29 |
frippe75 | tosky, ^^^ | 16:30 |
tosky | new deployment, new issues :) | 16:31 |
tosky | is it latest pike? Deployed with packstack too? | 16:31 |
frippe75 | Error: volumes_size: 0.0 is less than the minimum of 1 | 16:34 |
frippe75 | Could try switching from cinder type to epemeral | 16:34 |
frippe75 | ephemeral... | 16:34 |
frippe75 | Well I guess it could be initial pike. Deployed from pike-stable. | 16:34 |
frippe75 | I already stumbled on a couple of known issues.. But I was hoping stable meant basic nova/cinder/neutron functionality would be working. Not doing overly fancy things. | 16:35 |
tosky | but packstack or other ways? | 16:51 |
frippe75 | tosky, yes packstack | 17:06 |
frippe75 | trying to make some changes to my templates. removed them from admin as public and creating them in my user project instead. | 17:07 |
frippe75 | noticed storagelocation is cinder and volume type is None and not iscsi... | 17:08 |
frippe75 | tusky, not sure but my instances is now in the running state. Let's see if the deploy is a success... | 17:11 |
frippe75 | Think since the error was related to cinder. The template has gotten some new attributes I guess since I create my script (creating the json). Will fire up my old newton install and have a look. | 17:17 |
frippe75 | tosky, do I need to set both is_proxy_gateway=true and assign a floating ip ? | 18:04 |
frippe75 | ... for the "Wait for instance accessibility" ? | 18:06 |
frippe75 | I set the gateway but not the floating ip. If floating is mandatory for when using the gateway proxy I guess it could be controlled in the ui. | 18:07 |
frippe75 | I guess the answer was yes. | 18:12 |
frippe75 | in state "preparing" | 18:12 |
tosky | frippe75: if is_proxy_gateway is true for some instance, sahara will use it to reach the others | 18:21 |
tosky | so no, the floating ip it's not needed in that case | 18:21 |
openstackgerrit | Jeremy Freudberg proposed openstack/sahara-image-elements master: Revise s3_hadoop https://review.openstack.org/522990 | 18:21 |
tosky | but you need to ensure that the proxy command (that can be overridden through config) is working properly | 18:21 |
frippe75 | Tosky, well It didn't work without assigning the floating ip to the master as well as tagging it as a proxy gateway. | 18:23 |
frippe75 | Tried without floating and cluster could be reached as part of step3 | 18:24 |
frippe75 | tosky, Yes! now status is green and the cluster got fully deployed. My old cluster node templates was not complete to be used on pike. | 18:25 |
tosky | frippe75: uhm, I don't remember major changes from newton (and ocata) to pike that would make that not working | 18:39 |
tosky | frippe75: oh, I misunderstood: do you mean that you added the floating ip only to the node that it's the proxy gateway? | 18:40 |
openstackgerrit | Jeremy Freudberg proposed openstack/sahara-image-elements master: Revise s3_hadoop https://review.openstack.org/522990 | 18:49 |
*** tesseract has quit IRC | 18:50 | |
frippe75 | tosky, exactly only floating ip for the master node. works now, thanks for chatting! | 18:54 |
tosky | then yes: the proxy node must be reachable from the (sahara) controller | 19:12 |
tosky | happy to hear that it's working! | 19:12 |
*** dave-mccowan has joined #openstack-sahara | 19:27 | |
openstackgerrit | Jeremy Freudberg proposed openstack/sahara-image-elements master: Revise s3_hadoop https://review.openstack.org/522990 | 20:07 |
*** tosky has quit IRC | 20:47 | |
*** tosky has joined #openstack-sahara | 20:50 | |
*** ukaynar has quit IRC | 22:54 | |
*** ukaynar has joined #openstack-sahara | 22:59 | |
*** frippe75 has quit IRC | 23:09 | |
*** rcernin has joined #openstack-sahara | 23:15 | |
*** ukaynar has quit IRC | 23:17 | |
*** ukaynar has joined #openstack-sahara | 23:17 | |
*** ukaynar_ has joined #openstack-sahara | 23:21 | |
*** ukaynar has quit IRC | 23:23 | |
*** ukaynar_ has quit IRC | 23:26 | |
*** tellesnobrega has quit IRC | 23:38 | |
*** tellesnobrega has joined #openstack-sahara | 23:40 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!