| *** mhen_ is now known as mhen | 01:16 | |
| sylvr | Good Morning ! | 07:26 |
|---|---|---|
| sylvr | I think I might have a timeout issue on cinder volume creation : "Build of instance [...] aborted: Volume [...] did not finish being created even after we waited 194 seconds or 61 attempts. And its status is downloading." | 07:26 |
| sylvr | Looking at the state of the volume after the error and the volume exist without error, looking at the logs on the storage host show no error. | 07:26 |
| sylvr | Image download 559.88 MB at 74.81 MB/s and Converted 3584.00 MB image at 10.29 MB/s seems a bit slow but I don't have any baseline to compare too. | 07:26 |
| sylvr | Could you tell me if I'm on the right tracks and what should I do to fix this ? Thank you ! :) | 07:26 |
| Paul_ | Hi Folks | 10:46 |
| Paul_ | I've setup an Openstack cluster using Octavia for Load balancing | 10:46 |
| Paul_ | Does anyone know if the ratio to listeners to Load Balancers matter? | 10:46 |
| Paul_ | i.e. I could have a 100 services using 100 listeners on a single LB, versus creating a 100 Load Balancers | 10:47 |
| Paul_ | Does anyone have any experience with this or can give any advice about pros/cons? | 10:47 |
| Mohamed_Selim | There is no recommended ration for number of listeners per LB, but but in you consideration that listener shouldn't has conflict in port/ip, also like normal machine, more listener per LB, more cpu and memory load to distribute traffic | 12:39 |
| Mohamed_Selim | There is no recommended ration for number of listeners per LB, but, make in you consideration that listener shouldn't has conflict in port/ip, also like normal machine, more listener per LB, more cpu and memory load to distribute traffic | 12:39 |
| lowercase | I'm currently troubleshooting this issue: Live Migration failure: operation failed: guest CPU doesn't match specification: missing features: hle,rtm,rsba: libvirt.libvirtError: operation failed: guest CPU doesn't match specification: missing features: hle,rtm,rsba | 14:34 |
| lowercase | I can confirm 6 of the 8 instances in the host aggregate are indeed hle and rtm capable. No hypervisors in the HA are rsba compatible | 14:35 |
| lowercase | Performing a virsh dumpxml, I can confirm `<feature policy='require' name='rsba'/>` is present which is odd. | 14:36 |
| lowercase | But also, <feature policy='require' name='amd-stibp'/> | 14:36 |
| lowercase | Where all the servers in this HA are all intel | 14:36 |
| lowercase | All hypervisors are ubuntu 20, with libvirt 8. Openstack version 2023.1 | 14:37 |
| lowercase | Now, what i am curious is there is a diff HA with amd servers in it. They were added recently and I'm wondering if that is related. | 14:40 |
| DHE | migration requires the CPU type match... if you aren't performing feature leveling, you can't migrate between different types of CPUs. No OS supports calls to "cpuid" to return different data on different calls. | 14:46 |
| DHE | once an OS has been booted supporting some CPU instruction, it can only be migrated to another CPU supporting that same instruction. but you can have instructions disabled to provide a lowest-common-denominator sort of thing.. | 14:47 |
| lowercase | DHE: thanks for taking the time to respond. All servers in this server aggregate are `Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz`. I just double checked that right now and can confirm, all cpu's in this HA are identical. | 14:50 |
| lowercase | DHE: found the issue. Turns out, the Host Aggregates were not properly configured and the AZ spanned two different HEs, and when a live migration was being attempted, nova scheduler would send vms to the TSX disabled hypervisors. | 19:00 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!