Earlier  
Posted Nick Remark
#openstack-nova - 2018-05-17
16:19:45 mriedem so powervm was using 128 in that case
16:20:02 efried okay.
16:21:04 melwitt okay, so increasing the max is cool, it might be that 1024 is too high
16:21:14 mriedem MEDIUMTEXT [CHARACTER SET charset_name] [COLLATE collation_name] A TEXT column with a maximum length of 16,777,215 (224 − 1) characters. The effective maximum length is less if the value contains multibyte characters. Each MEDIUMTEXT value is stored using a 3-byte length prefix that indicates the number of
16:21:15 mriedem bytes in the value.
16:21:29 mriedem idk
16:21:34 mriedem someone should test it
16:21:49 mriedem i don't want to guess wtf would happen if someone tries to boot from volume with 1024 volumes
16:22:05 mriedem Tsuyoshi Nagata should test that
16:22:11 melwitt I'll ask
16:22:31 mriedem what is even the practical application of this?
16:23:44 openstackgerrit Merged openstack/nova master: Skip ServerActionsTestJSON.test_rebuild_server for cells v1 job https://review.openstack.org/567256
16:23:51 openstackgerrit Merged openstack/nova master: Remove mox in libvirt/test_driver.py (1) https://review.openstack.org/566555
16:23:57 openstackgerrit Merged openstack/nova stable/pike: libvirt: Report the virtual size of RAW disks https://review.openstack.org/568369
16:24:02 dansmith mriedem: I bet it is "because kvm can do it and nova can't"
16:24:04 openstackgerrit Merged openstack/nova master: Remove mox in test_live_migrate.py https://review.openstack.org/564656
16:24:05 dansmith not for any real reason
16:24:12 openstackgerrit Merged openstack/nova master: PowerVM Driver: DiskAdapter parent class https://review.openstack.org/549053
16:24:13 mriedem NFV something or other
16:24:16 dansmith or "because kvm can do it and nova won't let you"
16:24:38 mriedem in today's modern high-demand enterprise telco world, 26 volumes is for babies
16:24:52 melwitt I had thought it was something to do with people attaching a lot of volumes for HA or something?
16:25:02 mriedem that's a hell of a lot of HA :)
16:25:14 melwitt I remember it was mentioned in the nova/cinder session at the PTG in the context of the bulk volume create/attach discussion
16:25:17 mriedem this needs to be....super fucking redundant
16:25:28 mriedem all i took out of that discussion was,
16:25:47 mriedem "the storage backend allows it, so we want to make the software do it"
16:26:10 mriedem "because the admin can attach 500 volumes in 2 hours rather than 10, but they'll never touch this thing again after that"
16:26:34 mriedem so i'm kind of -1 on this blueprint just based on the "why?"
16:26:50 dansmith yar
16:27:50 mriedem where is ken'ichi when i want to praise him
16:28:08 melwitt okay. I thought it had been discussed in the past that some people attach a lot of volumes for an HA use case
16:28:31 mriedem idk maybe, that's certainly an argument for volume multiattach
16:28:42 mriedem but that's not attaching >1 volume to an instance
16:28:46 melwitt and maybe 1024 is too high but I was thinking it didn't sound unreasonable to increase it > 26 for such a use case
16:28:50 mriedem and i don't see how 1024 is sane HA
16:29:02 mriedem sure, i mean maybe 128 like the powervm thing
16:29:10 mriedem but need to actually justify why we should make this change
16:29:12 mriedem not just because we can
16:29:21 melwitt yeah, I think the point of the blueprint is > 26 and we were trying to find a reasonable max
16:29:42 mriedem sure, that's just not the first question i'd ask
16:30:06 mriedem i want shelve to send me an email when it's done
16:30:15 mriedem not because i need it, just because i can make it happen
16:30:25 mriedem i actually think we've had that blueprint...the email thing anyway
16:30:39 melwitt okay, considering we've raised it to 128 for powervm and 256 for virtio-scsi, I didn't think we needed to ask about > 26 for this case as to why. others are attaching a lot more devices than 26
16:31:55 dansmith mriedem: melwitt one way to look at it is 1024 is probably more than you could expect reasonable throughput from any of them on, given the number of host threads you'd need for them
16:33:12 melwitt sure ... I'm sorry I picked 1024, I didn't know it would be way too high. they didn't ask for 1024, they actually started with MAX_INT and I wanted to lower it to something more reasonable
16:34:32 dansmith tbh, I don't think it matters that much.. if it's not a big change, then whatever, but the justification can't be "because this is a legit real-world use case"
16:38:36 melwitt okay, I'll ask them to lower the proposed max and re-ask how many volumes they need to attach for their use case
16:41:57 edleafe dansmith: your change https://review.openstack.org/#/c/569127/ is failing unit tests due to the one-off change. Want me to fix it?
16:42:26 dansmith edleafe: ah, I figured it would.. I can do it
16:42:51 edleafe dansmith: ok
16:42:51 dansmith glad we have coverage on that
16:45:57 belmorei_ dansmith: me and tssurya were debugging the scheduler time. It's related with https://github.com/openstack/nova/blob/master/nova/scheduler/host_manager.py#L772
16:46:20 belmorei_ creating a dict with all instances info per host
16:48:01 dansmith per host we're considering each time we run through yeah?
16:48:14 openstackgerrit Eric Berglund proposed openstack/nova master: PowerVM Driver: Localdisk https://review.openstack.org/549300
16:48:20 belmorei_ yeah
16:49:21 dansmith belmorei_: what is the slow part exactly? it looks like it's hitting a cell db once per host instead of grabbing a bunch at a time maybe?
16:49:46 melwitt I've actually seen that complaint before and I don't understand how compute isn't sending updates to the scheduler. or does that not happen anymore because of placement?
16:49:56 mriedem it does
16:49:58 dansmith melwitt: that would be an upcall
16:49:59 mriedem if configured to do so
16:50:04 dansmith right
16:50:05 melwitt I see
16:50:09 mriedem it's an rpc cast from all computes to the scheduler topic
16:50:14 mriedem it's only used for the affinity filters
16:50:29 mriedem track_instance_changes or something like that
16:50:44 dansmith yeah, and what I'm not getting is why this would be different for them in the new arrangement,m
16:50:47 mriedem https://docs.openstack.org/nova/latest/configuration/config.html#filter_scheduler.track_instance_changes
16:50:55 dansmith because they're only considering hosts from one cell at this point anyway
16:50:56 mriedem dansmith: i'm wondering the same
16:51:01 melwitt yeah. lemme see if I can find the bug I'm thinking of
16:51:51 bauzas is someone able to tell me how to simply just associate an instance and a floating IP, without going thru the list of ports ?
16:52:07 bauzas we had that in the past, but we removed that here
16:52:12 melwitt this one https://bugs.launchpad.net/nova/+bug/1737465
16:52:13 openstack Launchpad bug 1737465 in OpenStack Compute (nova) "[cellv2] the performance issue of cellv2 when creating 500 instances concurrently" [Undecided,Confirmed] - Assigned to Jiang (jiangpf)
16:52:20 belmorei_ dansmith not following... hosts from one cell?
16:52:28 mriedem bauzas: I CAN
16:52:39 mriedem bauzas: https://docs.openstack.org/python-openstackclient/pike/cli/command-objects/floating-ip.html#floating-ip-set
16:52:45 mriedem get the port id from the instance
16:52:51 mriedem then associate the port to hte floating ip
16:52:55 bauzas mriedem: port is required
16:52:55 mriedem or create the floating ip with the port
16:52:57 dansmith belmorei_: that code should be building a dict of host info, but only from the list of uuids that we got back from placement, which should be only one cell, and only the subset of those that are candidates
16:53:05 mriedem bauzas: yes the port is what links the instance to the floating ip
16:53:07 bauzas mriedem: so I need to openstack port list before
16:53:23 bauzas I was thinking we could somehow magically avoid one roundtrip
16:53:33 bauzas with openstack CLI doing that in secret for me
16:53:38 mriedem you can do it on a single command line if you want...
16:53:46 bauzas awk my love
16:53:50 mriedem no
16:53:51 mriedem NO
16:54:01 mriedem PORT1=$(openstack port list --device-id $SERVER1 -c ID -f value)
16:54:07 mriedem FLOATINGIP1=$(openstack floating ip create --port $PORT1 public -c floating_ip_address -f value)
16:54:13 mriedem do that in a single line
16:54:22 bauzas meh
16:54:23 belmorei_ dansmith: yes, but takes a lot of time even for 200 nodes
16:54:24 mriedem then $$$ me
16:54:34 mriedem bauzas: meh?

Earlier   Later