| opendevreview | Merged openstack/openstack-ansible-os_designate master: Add quorum queues support for service https://review.opendev.org/c/openstack/openstack-ansible-os_designate/+/898818 | 10:03 |
|---|---|---|
| opendevreview | Andrew Bonney proposed openstack/openstack-ansible-os_ironic master: Use common value for inspector callback URL https://review.opendev.org/c/openstack/openstack-ansible-os_ironic/+/899944 | 10:48 |
| anskiy | I don't see quorum queue support in keystone, is this intended? | 12:07 |
| jrosser | that could be a mistake i guess | 12:22 |
| jrosser | there are notifications from keystone | 12:22 |
| anskiy | and it's vhost is still prefixed with `/`, like non-quorum ones now :) | 12:29 |
| anskiy | oh, noonedeadpunk's not here this week, I'll try to remember to ask this thing later | 12:33 |
| NeilHanlon | jrosser: I was going to give my +2 on the rsyslog server/client changes. Should I also give it a W+? | 13:06 |
| NeilHanlon | just wanna make sure it's good to merge :) | 13:06 |
| anskiy | seems like os_swift does produce some notifications too, at least there is a reference in proxy-server.conf | 13:17 |
| NeilHanlon | fyi https://review.opendev.org/q/topic:osa%252Fquorum_queues | 13:19 |
| anskiy | NeilHanlon: yeah, I've just checked that neither os_swift, nor os_keystone has outstanding changes for quorum queues | 13:21 |
| spatel | anyone running AMD processor here? | 13:39 |
| NeilHanlon | i have one on my laptop | 13:39 |
| NeilHanlon | but not in prod really -- aside from AWS | 13:39 |
| spatel | we bought brand new supermicro server - AMD EPYC 7302 16-Core Processor | 13:40 |
| spatel | Deploy ubuntu 22.04 and sucker running super slow :( | 13:40 |
| spatel | No error.. nothing | 13:40 |
| spatel | Just slow.. | 13:40 |
| spatel | This is latest CPU with 64 cores | 13:40 |
| spatel | I am running it BIOS legacy mode.. hope it has nothing to do with UEFI mode | 13:41 |
| mgariepy | i run amd cpus with efi not issue on 22.04 | 13:41 |
| NeilHanlon | also, not that it should be slow, but that _is_ a four year old processor, still | 13:42 |
| NeilHanlon | how are thermals on the box? | 13:42 |
| NeilHanlon | maybe it's throttling? | 13:42 |
| * NeilHanlon just had a customer with a "slow" HP server that was fixed by regooping the cpu heatsinks | 13:43 | |
| spatel | mgariepy you are using EFI? | 13:44 |
| mgariepy | i've seen slow compute nodes when ram was Error correcting like crazy | 13:44 |
| mgariepy | yes | 13:44 |
| spatel | After rebooting server its normal now.. but yesterday it was super crazy slow | 13:44 |
| spatel | I have 3 servers.. and they are brand new.. just unbox yesterday | 13:45 |
| spatel | 256GB memory with Intel NVME | 13:45 |
| NeilHanlon | i'd run memtest and sysbench on it, honestly. sometimes you just get a bad one :\ | 13:46 |
| spatel | Does EFI matter (I thought its just to load grub and nothing else) | 13:46 |
| NeilHanlon | I'd _recommend_ EFI, but I don't know it matters per se | 13:46 |
| spatel | Do I need to re-install ubuntu if I switch on EFI? | 13:46 |
| spatel | I never use EFI | 13:46 |
| mgariepy | why ? | 13:47 |
| NeilHanlon | basically, yes... it's _possible_ to convert over, but, it's probably not worth the effort. You need a specificly-formatted /boot/efi for it to work | 13:47 |
| NeilHanlon | (among other things) | 13:47 |
| spatel | Do i need to download UEI supported Ubuntu ISO or its same ISO? | 13:48 |
| NeilHanlon | same iso afaik | 13:48 |
| anskiy | you don't actually need grub with EFI and I wonder, why distros still install it with EFI | 13:48 |
| mgariepy | bios mode is legacy | 13:48 |
| spatel | I have BIOS mode legacy currently | 13:49 |
| NeilHanlon | anskiy: I guess you _could_ load Linux right from shim.. but.. | 13:49 |
| anskiy | the only thing is that you actually need to check if you can boot from those NVMes: if they are shown in efishell | 13:49 |
| anskiy | NeilHanlon: that's what I'm running right now on my laptop :) | 13:49 |
| spatel | Hmm | 13:50 |
| anskiy | spatel: besides, what does "slow" mean? :) | 13:50 |
| NeilHanlon | anskiy: not fast! ;) | 13:50 |
| spatel | when I do 'apt install foo' then it will stuck at 40% or some random % for almost 15 min... | 13:51 |
| NeilHanlon | also, spatel, this is off topic but. are you in Massachusetts? (I ask because that's where I live...) | 13:51 |
| spatel | CPU go high | 13:51 |
| spatel | no errors in logs.. no dmesg error | 13:51 |
| spatel | I am in MA :) | 13:51 |
| spatel | where are you from in MA (send me DM if its personal) | 13:52 |
| * NeilHanlon saw your irc host last time you parted :P | 13:52 | |
| NeilHanlon | I'm in Bedford | 13:52 |
| spatel | Boy..not far from my place | 13:52 |
| spatel | I am near framingham | 13:52 |
| anskiy | spatel: well... strace? ;) | 13:52 |
| spatel | Will catch sometime for beer then :) | 13:53 |
| spatel | anskiy after frustration I just reboot server and that fixed it.. but that happened to all 3 servers and same time | 13:53 |
| spatel | They are my new cloud controller :( | 13:54 |
| NeilHanlon | spatel: totally :) | 13:54 |
| spatel | I am very curious how that happened on all 3 server and same time | 13:54 |
| spatel | NeilHanlon +1 | 13:54 |
| NeilHanlon | it's odd that they would all act that way.. did you happen to see what the nodes were doing while the CPU was high? any particular process stand out? | 13:54 |
| mgariepy | might have a issue with the command they were running.. if they reacted the same way on the apt update on the same package. | 13:55 |
| spatel | I noticed bunch of zombie process and specially on fwupdmgr daemon | 13:56 |
| NeilHanlon | hmm | 13:56 |
| spatel | I am wondering what if fwupdmgr was doing something nasty.. | 13:56 |
| NeilHanlon | you know, that could explain it.. if it was updating firmwares | 13:56 |
| NeilHanlon | both the slowness *and* them being 'better' on reboot | 13:56 |
| spatel | I am trying to run fwupdmgr again manually and see if that can re-create issue | 13:57 |
| NeilHanlon | I'd check out what fwupd did -- maybe look at errata for the Supermicro box and see what firmwares they've put out | 13:57 |
| spatel | when I run that utility I got this warning - WARNING: UEFI firmware can not be updated in legacy BIOS mode | 13:58 |
| mgariepy | how to you install the server ? via curtain ? | 13:59 |
| spatel | via IOS with mounting disk on BMC | 13:59 |
| spatel | ISO* | 13:59 |
| mgariepy | manual install ? | 13:59 |
| spatel | yes | 14:00 |
| mgariepy | ok i don't like doing manual installs :D too error prone for me. | 14:02 |
| spatel | I have only 3 server so thought not worth setup PXE etc.. | 14:02 |
| mgariepy | write an ansible playbook to deploy it for you ;p then it's easier to deploy via pxe than not :) | 14:03 |
| spatel | :) | 14:03 |
| spatel | Let me update firmware and see.. if issue come back then I will turn UEFI | 14:04 |
| opendevreview | Danila Balagansky proposed openstack/openstack-ansible-os_ceilometer master: Convert role_name to list https://review.opendev.org/c/openstack/openstack-ansible-os_ceilometer/+/899970 | 15:36 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!