Perl 6 - the future is here, just unevenly distributed

IRC log for #fuel, 2015-01-16

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary

All times shown according to UTC.

Time Nick Message
00:25 teran joined #fuel
00:53 emagana joined #fuel
00:59 angryjesters joined #fuel
01:21 xarses joined #fuel
01:36 rmoe joined #fuel
01:51 virtuallynathan_ joined #fuel
02:45 brad[] joined #fuel
03:08 Longgeek joined #fuel
03:12 Obi-Wan joined #fuel
03:42 Longgeek joined #fuel
03:49 jobewan joined #fuel
04:07 Longgeek joined #fuel
04:13 teran joined #fuel
04:21 fandi_ joined #fuel
06:08 sambork joined #fuel
06:16 saibarspeis joined #fuel
06:16 youellet joined #fuel
06:31 taj joined #fuel
06:33 saibarspeis joined #fuel
06:36 jobewan joined #fuel
06:46 robklg joined #fuel
06:48 jobewan joined #fuel
07:02 teran joined #fuel
07:20 Longgeek joined #fuel
07:25 dklepikov joined #fuel
07:36 teran joined #fuel
07:47 sressot joined #fuel
07:52 Miouge joined #fuel
07:57 dkusidlo joined #fuel
08:04 corepb_ joined #fuel
08:10 getup joined #fuel
08:29 teran joined #fuel
08:30 teran_ joined #fuel
08:31 teran_ joined #fuel
08:31 teran_ joined #fuel
08:38 monester_laptop joined #fuel
08:40 mescanef joined #fuel
08:48 HeOS joined #fuel
08:49 monester_laptop joined #fuel
09:01 devstok joined #fuel
09:02 devstok hi all
09:02 devstok talking on CEPH
09:02 devstok replica factor
09:02 avgoor joined #fuel
09:06 devstok the replica factor can slow the performance of entire system ?
09:06 devstok replica 3 is slower than replica 2 or I can't see any difference?
09:09 samuelbartel joined #fuel
09:13 pawels joined #fuel
09:14 ignatenko joined #fuel
09:15 tatyana joined #fuel
09:18 robklg joined #fuel
09:28 pawels1 joined #fuel
09:31 dkusidlo joined #fuel
09:37 monester_laptop joined #fuel
09:45 aliemieshko_ joined #fuel
09:47 igajsin-home joined #fuel
09:48 Longgeek joined #fuel
09:52 Longgeek_ joined #fuel
09:56 Longgeek joined #fuel
09:57 dkusidlo joined #fuel
10:01 stamak joined #fuel
10:03 evg devstok: It can impact performance at least indirectly making replication in background. Replica factor 3 is the minimum recomended for production env.
10:05 devstok can it impacts on the creation of VM ?
10:06 devstok because , I've deployed the same env 2 times and the only difference was replica factor
10:06 e0ne joined #fuel
10:07 devstok with factor replica 3 the vm created with qcow2 was spawned in 15 miuntes, with replica factor 2 the vm with qcow2 was spawned in 2 minutes
10:25 Longgeek_ joined #fuel
10:27 Longgeek_ joined #fuel
10:36 Longgeek joined #fuel
10:40 geekinutah joined #fuel
10:48 evg devstok: you should use raw format in case of ceph for better performance. but 7 times degradation is something strange anyway.
10:49 devstok yes strange but the only difference between the two env was the replica factor
10:49 devstok best performance with replica 2
10:53 artem_panchenko left #fuel
10:56 pawels joined #fuel
10:59 teran joined #fuel
11:03 pawels joined #fuel
11:04 teran_ joined #fuel
11:12 tobiash joined #fuel
11:16 e0ne joined #fuel
11:28 evg devstok: it obviously must be slightly better. I read some a benchmark about the case, will trey to find url.
11:28 dkusidlo joined #fuel
11:30 ddmitriev joined #fuel
11:46 fandi joined #fuel
11:48 Arminder joined #fuel
11:49 tobiash_ joined #fuel
11:52 Arminder- joined #fuel
11:57 Arminder joined #fuel
11:59 Arminder- joined #fuel
12:01 robklg joined #fuel
12:23 sc-rm evg: If I expirencese slow write performance on our compute instaces using ceph as backend, is it then the ceph cluster or the compute node that might be the bottleneck?
12:25 Arminder joined #fuel
12:26 monester_laptop joined #fuel
12:26 robklg joined #fuel
12:29 Longgeek joined #fuel
12:31 saibarspeis joined #fuel
12:32 MiroslavAnashkin It may be either ceph or network performance.
12:53 dkusidlo joined #fuel
13:04 monester_laptop joined #fuel
13:05 dkusidlo joined #fuel
13:10 Longgeek joined #fuel
13:28 emagana joined #fuel
13:35 jaypipes joined #fuel
13:42 miroslav_ joined #fuel
13:49 robklg joined #fuel
13:54 championofcyrod1 I may have found a bug with the Murano health check... how do I know if already reported?
13:55 MiroslavAnashkin Please search https://bugs.launchpad.net/murano
13:55 championofcyrod1 basically... the murano health check uses the same environment table name.
13:55 championofcyrod1 http://paste.openstack.org/show/158249/
13:55 championofcyrod1 so if it failed at one point.  (and you fix rabbitmq like you mentioned to me yesterday)... then re-running the health check continually fails until you remove the record.
13:56 championofcyrod1 like so: http://paste.openstack.org/show/158253/
13:56 Miouge_ joined #fuel
13:56 championofcyrod1 is the 'health check' in fuel part of murano or fuel?
13:59 fandi joined #fuel
14:07 fandi joined #fuel
14:15 fandi joined #fuel
14:15 benrodrigue joined #fuel
14:15 MiroslavAnashkin There are several health checks. This one is from Murano. Murano health check needs some additional preparatuin steps, if I remember correct
14:17 championofcyrod1 sorry. I wasnt clear.  is the 'murano' health check performed by making several requests to the murano api? or is there a murano api function that just does a 'test' ?
14:17 monester_laptop joined #fuel
14:17 championofcyrod1 if it was just a murano 'test' api call.. then the issue would be murano... no?
14:23 vsedelnik joined #fuel
14:23 vsedelnik left #fuel
14:41 devstok HEY!
14:42 devstok How ceph decide where to deploy the replica? If I set replica at 2 may I have the replica in the same host with 2 OSD ?
14:47 tatyana Hi championofcyrod1, you are absolutelly righ, murano' health check performed by making several requests to the murano api. So here http://paste.openstack.org/show/158249/  seems the health check issue.  Could you please register the issue about this https://bugs.launchpad.net/fuel here with tag ostf and with snapshot
14:53 championofcyrod1 tatyana: sure. i have a launchpad account.
14:57 pawels1 joined #fuel
15:00 championofcyrod1 tatyana: done. https://bugs.launchpad.net/fuel/+bug/1411701
15:00 tatyana thank you championofcyrod1 :)
15:00 emagana joined #fuel
15:02 mattgriffin joined #fuel
15:03 championofcyrod1 So now I have another issue w/ my fuel stack...  and it's a real problem.  Functional tests for creating volumes and booting instance from it, as well as creating volume and attaching it to an instance is failing.  storage is ceph.  the fuel logs indicate that there was a time waiting for active status from the instance for the first, and then errored at attaching volume to an instance for the second.
15:03 championofcyrod1 but i'm not finding any cinder or glance errors relevant.
15:03 championofcyrod1 and the /var/log/ceph/ceph.log was not helpful either.
15:03 mattgri__ joined #fuel
15:05 teran joined #fuel
15:06 championofcyrod1 captured log is huge: http://paste.openstack.org/show/158264/
15:14 SamYaple [A4
15:14 SamYaple ^^ ignore
15:20 Samos123 do compute nodes still talk over the management network to ceph? according to Fuel + ceph slides that was how it's previously done, so if VM wants fastest bandwidth to ceph you would have to make the management network 10Gb? is it still like that?
15:31 MiroslavAnashkin Yes, VMs talk with Cinder - which is in Management network.
15:32 fandi joined #fuel
15:32 evg Samos123: It's still so. See https://blueprints.launchpad.net/fuel/+spec/fuel-storage-networks
15:33 blahRus joined #fuel
15:36 fandi joined #fuel
15:40 fandi joined #fuel
15:40 Samos123 thanks for answer and related bp!
15:44 fandi joined #fuel
15:44 Samos123 so if you're using cinder + lvm the storage network wouldn't be used at all?
15:45 e0ne joined #fuel
15:45 emagana joined #fuel
15:49 devstok can I manually change  the replica factor ?
15:52 jobewan joined #fuel
15:56 emagana joined #fuel
15:58 pawels joined #fuel
15:58 pawels left #fuel
16:04 emagana joined #fuel
16:07 sambork joined #fuel
16:14 teran joined #fuel
16:20 sambork joined #fuel
16:21 Miouge joined #fuel
16:23 jobewan joined #fuel
16:33 MiroslavAnashkin Yes, storage is to sync storage devices
16:35 MiroslavAnashkin devstok: yes. BTW, it is per data pool setting `ceph osd pool set {poolname} size {num-replicas}`
16:36 MiroslavAnashkin http://ceph.com/docs/master/rados/operations/pools/
16:40 devstok thanks
16:45 Longgeek joined #fuel
16:49 benrodrigue joined #fuel
16:55 Longgeek joined #fuel
17:00 emagana joined #fuel
17:03 emagana_ joined #fuel
17:15 rmoe joined #fuel
17:19 mutex joined #fuel
17:30 ignatenko left #fuel
17:35 Arminder- joined #fuel
17:38 teran_ joined #fuel
17:40 Arminder joined #fuel
17:58 thumpba joined #fuel
18:17 Miouge joined #fuel
18:28 e0ne joined #fuel
18:30 teran joined #fuel
18:44 xarses joined #fuel
18:55 thumpba joined #fuel
19:07 HeOS joined #fuel
19:09 mpetason joined #fuel
19:16 fandi joined #fuel
19:23 igajsin-home left #fuel
19:27 Arminder joined #fuel
19:30 Arminder joined #fuel
19:30 thumpba joined #fuel
19:31 thumpba joined #fuel
19:34 Arminder joined #fuel
19:55 Arminder- joined #fuel
20:29 Arminder joined #fuel
20:31 Arminder joined #fuel
20:43 Longgeek joined #fuel
21:03 emagana joined #fuel
21:05 emagana joined #fuel
21:06 Longgeek_ joined #fuel
21:14 emagana joined #fuel
21:24 stamak joined #fuel
21:25 mattgriffin joined #fuel
21:28 e0ne joined #fuel
21:44 fandi joined #fuel
22:02 youellet joined #fuel
22:07 mpetason joined #fuel
22:45 ddmitriev joined #fuel
23:04 emagana joined #fuel
23:11 emagana joined #fuel
23:27 jetpks joined #fuel
23:40 emagana joined #fuel
23:58 thumpba joined #fuel
23:58 thumpba joined #fuel

| Channels | #fuel index | Today | | Search | Google Search | Plain-Text | summary