Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

scylla_prepare: generating wrong perftune.yaml that could not run clocksource setting #7672

Closed
syuu1228 opened this issue Nov 22, 2020 · 1 comment
Assignees
Labels
cloud/aws AWS related issues
Milestone

Comments

@syuu1228
Copy link
Contributor

This is Scylla's bug tracker, to be used for reporting bugs only.
If you have a question about Scylla, and not a bug, please ask it in
our mailing-list at scylladb-dev@googlegroups.com or in our slack channel.

  • [] I have read the disclaimer above, and I am reporting a suspected malfunction in Scylla.

Installation details
Scylla version (or git commit hash): 1e170eb
Cluster size:
OS (RHEL/CentOS/Ubuntu/AWS AMI): Amazon Linux 2

Hardware details (for performance issues) Delete if unneeded
Platform (physical/VM/cloud instance type/docker): EC2 i3en.3xlarge
Hardware: sockets= cores= hyperthreading= memory=
Disks: (SSD/HDD, count)

It seems like setting SET_CLOCKSOURCE=yes on /etc/sysconfig/scylla-server does not enables clocksource setup.
While scylla_prepare running, perftune.py should say Setting clocksource to kvm-clock if clocksource setup enabled, but I don't find it.
Also current_clocksource does not changed.
I suspect it's perftune.yaml problem, it missing tune: - system.
After I added it, it just works finely.

[scyllaadm@ip-10-0-0-75 ~]$ grep CLOCKSOURCE /etc/sysconfig/scylla-server 
SET_CLOCKSOURCE="yes"
[scyllaadm@ip-10-0-0-75 ~]$ cat /sys/devices/system/clocksource/clocksource0/current_clocksource 
tsc
[scyllaadm@ip-10-0-0-75 ~]$ sudo /opt/scylladb/scripts/scylla_prepare
File /etc/sysconfig/irqbalance.scylla.orig already exists - not overwriting.
Restarting irqbalance: going to ban the following IRQ numbers: 24, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 43, 45, 46, 50, 42, 48, 41, 51, 52, 44, 49, 47 ...
Restarting irqbalance via systemctl...
No non-NVMe disks to tune
Setting NVMe disks: nvme1n1...
Setting mask 00000001 in /proc/irq/24/smp_affinity
Writing 'none' to /sys/devices/pci0000:00/0000:00:1f.0/nvme/nvme1/nvme1n1/queue/scheduler
Writing '2' to /sys/devices/pci0000:00/0000:00:1f.0/nvme/nvme1/nvme1n1/queue/nomerges
Setting a physical interface eth0...
Distributing all IRQs
Setting mask 00000001 in /proc/irq/43/smp_affinity
Setting mask 00000040 in /proc/irq/45/smp_affinity
Setting mask 00000002 in /proc/irq/46/smp_affinity
Setting mask 00000080 in /proc/irq/50/smp_affinity
Setting mask 00000004 in /proc/irq/42/smp_affinity
Setting mask 00000100 in /proc/irq/48/smp_affinity
Setting mask 00000008 in /proc/irq/41/smp_affinity
Setting mask 00000200 in /proc/irq/51/smp_affinity
Setting mask 00000010 in /proc/irq/52/smp_affinity
Setting mask 00000400 in /proc/irq/44/smp_affinity
Setting mask 00000020 in /proc/irq/49/smp_affinity
Setting mask 00000800 in /proc/irq/47/smp_affinity
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-9/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-11/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-7/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-5/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-3/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-1/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-8/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-10/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-6/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-4/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-2/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-0/rps_cpus
Setting net.core.rps_sock_flow_entries to 32768
Setting limit 2730 in /sys/class/net/eth0/queues/rx-9/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-11/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-7/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-5/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-3/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-1/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-8/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-10/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-6/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-4/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-2/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-0/rps_flow_cnt
Trying to enable ntuple filtering HW offload for eth0...not supported
Setting mask 00000001 in /sys/class/net/eth0/queues/tx-6/xps_cpus
Setting mask 00000040 in /sys/class/net/eth0/queues/tx-4/xps_cpus
Setting mask 00000002 in /sys/class/net/eth0/queues/tx-10/xps_cpus
Setting mask 00000080 in /sys/class/net/eth0/queues/tx-2/xps_cpus
Setting mask 00000004 in /sys/class/net/eth0/queues/tx-0/xps_cpus
Setting mask 00000100 in /sys/class/net/eth0/queues/tx-9/xps_cpus
Setting mask 00000008 in /sys/class/net/eth0/queues/tx-7/xps_cpus
Setting mask 00000200 in /sys/class/net/eth0/queues/tx-5/xps_cpus
Setting mask 00000010 in /sys/class/net/eth0/queues/tx-11/xps_cpus
Setting mask 00000400 in /sys/class/net/eth0/queues/tx-3/xps_cpus
Setting mask 00000020 in /sys/class/net/eth0/queues/tx-1/xps_cpus
Setting mask 00000800 in /sys/class/net/eth0/queues/tx-8/xps_cpus
Writing '4096' to /proc/sys/net/core/somaxconn
Writing '4096' to /proc/sys/net/ipv4/tcp_max_syn_backlog
[scyllaadm@ip-10-0-0-75 ~]$ cat /sys/devices/system/clocksource/clocksource0/current_clocksource 
tsc
[scyllaadm@ip-10-0-0-75 ~]$ cat /etc/scylla.d/perftune.yaml
cpu_mask: '0x00000fff'
mode: mq
nic: eth0
tune:
- net
tune_clock: true

After perftune.yaml changed:

[scyllaadm@ip-10-0-0-75 ~]$ cat /etc/scylla.d/perftune.yaml
cpu_mask: '0x00000fff'
mode: mq
nic: eth0
tune:
- net
- system
tune_clock: true
[scyllaadm@ip-10-0-0-75 ~]$ sudo /opt/scylladb/scripts/scylla_prepare
File /etc/sysconfig/irqbalance.scylla.orig already exists - not overwriting.
Restarting irqbalance: going to ban the following IRQ numbers: 42, 45, 49, 44, 50, 48, 43, 47, 52, 51, 46, 41 ...
Restarting irqbalance via systemctl...
Setting a physical interface eth0...
Distributing all IRQs
Setting mask 00000001 in /proc/irq/42/smp_affinity
Setting mask 00000040 in /proc/irq/45/smp_affinity
Setting mask 00000002 in /proc/irq/49/smp_affinity
Setting mask 00000080 in /proc/irq/44/smp_affinity
Setting mask 00000004 in /proc/irq/50/smp_affinity
Setting mask 00000100 in /proc/irq/48/smp_affinity
Setting mask 00000008 in /proc/irq/43/smp_affinity
Setting mask 00000200 in /proc/irq/47/smp_affinity
Setting mask 00000010 in /proc/irq/52/smp_affinity
Setting mask 00000400 in /proc/irq/51/smp_affinity
Setting mask 00000020 in /proc/irq/46/smp_affinity
Setting mask 00000800 in /proc/irq/41/smp_affinity
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-9/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-11/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-7/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-5/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-3/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-1/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-8/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-10/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-6/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-4/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-2/rps_cpus
Setting mask 00000fff in /sys/class/net/eth0/queues/rx-0/rps_cpus
Setting net.core.rps_sock_flow_entries to 32768
Setting limit 2730 in /sys/class/net/eth0/queues/rx-9/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-11/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-7/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-5/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-3/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-1/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-8/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-10/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-6/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-4/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-2/rps_flow_cnt
Setting limit 2730 in /sys/class/net/eth0/queues/rx-0/rps_flow_cnt
Trying to enable ntuple filtering HW offload for eth0...not supported
Setting mask 00000001 in /sys/class/net/eth0/queues/tx-6/xps_cpus
Setting mask 00000040 in /sys/class/net/eth0/queues/tx-4/xps_cpus
Setting mask 00000002 in /sys/class/net/eth0/queues/tx-10/xps_cpus
Setting mask 00000080 in /sys/class/net/eth0/queues/tx-2/xps_cpus
Setting mask 00000004 in /sys/class/net/eth0/queues/tx-0/xps_cpus
Setting mask 00000100 in /sys/class/net/eth0/queues/tx-9/xps_cpus
Setting mask 00000008 in /sys/class/net/eth0/queues/tx-7/xps_cpus
Setting mask 00000200 in /sys/class/net/eth0/queues/tx-5/xps_cpus
Setting mask 00000010 in /sys/class/net/eth0/queues/tx-11/xps_cpus
Setting mask 00000400 in /sys/class/net/eth0/queues/tx-3/xps_cpus
Setting mask 00000020 in /sys/class/net/eth0/queues/tx-1/xps_cpus
Setting mask 00000800 in /sys/class/net/eth0/queues/tx-8/xps_cpus
Writing '4096' to /proc/sys/net/core/somaxconn
Writing '4096' to /proc/sys/net/ipv4/tcp_max_syn_backlog
Setting clocksource to kvm-clock
[scyllaadm@ip-10-0-0-75 ~]$ cat /sys/devices/system/clocksource/clocksource0/current_clocksource 
kvm-clock
syuu1228 added a commit to syuu1228/scylla that referenced this issue Nov 22, 2020
perftune.py only run clocksource setup when --tune system specified,
so we need to add it on the parameter when SET_CLOCKSOURCE=yes.

Fixes scylladb#7672
@slivne slivne added this to the 4.3 milestone Nov 23, 2020
@slivne slivne added the cloud/aws AWS related issues label Nov 23, 2020
@avikivity
Copy link
Member

Fix present on all active branches, not backporting.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
cloud/aws AWS related issues
Projects
None yet
Development

No branches or pull requests

4 participants