Re: Seeing a lot of scale_rt_power messages
Linux on 390 Port LINUX-390@VM.MARIST.EDU wrote on 10/23/2013 10:46:35 AM: From: Richard Troth ri...@velocitysoftware.com To: LINUX-390@VM.MARIST.EDU, Date: 10/23/2013 10:46 AM Subject: Re: Seeing a lot of scale_rt_power messages Sent by: Linux on 390 Port LINUX-390@VM.MARIST.EDU I haven't seen it that I recall, but a Google search suggests that it comes from load balancing when you're running tickless. scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f The scheduler is trying to tell you something (because the scale_rt_power() function is in the scheduler), but the context is lost. Does this guest have multiple CPUs? 2 cpu's defined. We have other guests with multiple CPU's but don't exhibit this behavior. Also, have you made any tuning changes over its life? (Things handled by 'sysctl' or /etc/sysctl.conf.) Not that I am aware of. What is the output of ... sysctl -a | grep sched kernel.sched_child_runs_first = 0 kernel.sched_min_granularity_ns = 400 kernel.sched_latency_ns = 1200 kernel.sched_wakeup_granularity_ns = 500 kernel.sched_tunable_scaling = 1 kernel.sched_migration_cost = 50 kernel.sched_nr_migrate = 32 kernel.sched_time_avg = 1000 kernel.sched_shares_window = 1000 kernel.sched_rt_period_us = 100 kernel.sched_rt_runtime_us = 95 kernel.sched_compat_yield = 0 kernel.sched_cfs_bandwidth_slice_us = 5000 kernel.sched_domain.cpu0.domain0.min_interval = 1 kernel.sched_domain.cpu0.domain0.max_interval = 4 kernel.sched_domain.cpu0.domain0.busy_idx = 2 kernel.sched_domain.cpu0.domain0.idle_idx = 1 kernel.sched_domain.cpu0.domain0.newidle_idx = 0 kernel.sched_domain.cpu0.domain0.wake_idx = 0 kernel.sched_domain.cpu0.domain0.forkexec_idx = 0 kernel.sched_domain.cpu0.domain0.busy_factor = 64 kernel.sched_domain.cpu0.domain0.imbalance_pct = 125 kernel.sched_domain.cpu0.domain0.cache_nice_tries = 1 kernel.sched_domain.cpu0.domain0.flags = 4143 kernel.sched_domain.cpu0.domain0.name = CPU kernel.sched_domain.cpu1.domain0.min_interval = 1 kernel.sched_domain.cpu1.domain0.max_interval = 4 kernel.sched_domain.cpu1.domain0.busy_idx = 2 kernel.sched_domain.cpu1.domain0.idle_idx = 1 kernel.sched_domain.cpu1.domain0.newidle_idx = 0 kernel.sched_domain.cpu1.domain0.wake_idx = 0 kernel.sched_domain.cpu1.domain0.forkexec_idx = 0 kernel.sched_domain.cpu1.domain0.busy_factor = 64 kernel.sched_domain.cpu1.domain0.imbalance_pct = 125 kernel.sched_domain.cpu1.domain0.cache_nice_tries = 1 kernel.sched_domain.cpu1.domain0.flags = 4143 kernel.sched_domain.cpu1.domain0.name = CPU Also, what do your boot parms look like? (Look for HZ timer and other scheduler tweaks.) No parms on the boot for scheduler or HZ timer. -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Seeing a lot of scale_rt_power messages
Seeing a lot of these since we upgraded from SLES10SP4 to SLES11SP2: scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f Oct 23 04:43:19 sandbx3 kernel: scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f scale_rt_power: clock:3806a71c9a763 age:3806a6c2e6300, avg:2ed2b4f8 Oct 23 04:43:19 sandbx3 kernel: scale_rt_power: clock:3806a71c9a763 age:3806a6c2e6300, avg:2ed2b4f8 What's up with this? Any way to get around this? -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: Seeing a lot of scale_rt_power messages
I haven't seen it that I recall, but a Google search suggests that it comes from load balancing when you're running tickless. scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f The scheduler is trying to tell you something (because the scale_rt_power() function is in the scheduler), but the context is lost. Does this guest have multiple CPUs? Also, have you made any tuning changes over its life? (Things handled by 'sysctl' or /etc/sysctl.conf.) What is the output of ... sysctl -a | grep sched Also, what do your boot parms look like? (Look for HZ timer and other scheduler tweaks.) On Wed, Oct 23, 2013 at 7:32 AM, karlkings...@ongov.net wrote: Seeing a lot of these since we upgraded from SLES10SP4 to SLES11SP2: scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f Oct 23 04:43:19 sandbx3 kernel: scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f scale_rt_power: clock:3806a71c9a763 age:3806a6c2e6300, avg:2ed2b4f8 Oct 23 04:43:19 sandbx3 kernel: scale_rt_power: clock:3806a71c9a763 age:3806a6c2e6300, avg:2ed2b4f8 What's up with this? Any way to get around this? -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/ -- -- R; Rick Troth Velocity Software http://www.velocitysoftware.com/ -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/
Re: Seeing a lot of scale_rt_power messages
On 10/23/2013 at 07:32 AM, karlkings...@ongov.net wrote: Seeing a lot of these since we upgraded from SLES10SP4 to SLES11SP2: scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f Oct 23 04:43:19 sandbx3 kernel: scale_rt_power: clock:3806a691fbdb1 age:3806a4e60fe00, avg:5d4b8d2f scale_rt_power: clock:3806a71c9a763 age:3806a6c2e6300, avg:2ed2b4f8 Oct 23 04:43:19 sandbx3 kernel: scale_rt_power: clock:3806a71c9a763 age:3806a6c2e6300, avg:2ed2b4f8 What's up with this? Any way to get around this? You'll need to open up a service request with your service provider to get a really detailed answer. At a higher level, that message is coming out of kernel/sched_fair.c. The comment above it says: /* RT usage tracking looks fishy, report anomaly and restore sanity */ From what I can make out, RT refers to Real Time, but I don't see how that fits in with the Completely Fair Scheduler that you're apparently running. If I had to guess, I would say that you're seeing some sort of problem with the system clock not being right enough which could be due to some sort of resource starvation. Switching to the Deadline scheduler would probably make the messages go away (and I believe Deadline is the recommended scheduler for System z), but that wouldn't address the root cause. Mark Post -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390 -- For more information on Linux on System z, visit http://wiki.linuxvm.org/