Re: [PATCH 0/5] Alter steal time reporting in KVM
On 12/05/2012 06:46 AM, Glauber Costa wrote: I am deeply sorry. I was busy first time I read this, so I postponed answering and ended up forgetting. Sorry include/linux/sched.h: unsigned long long run_delay; /* time spent waiting on a runqueue */ So if you are out of the runqueue, you won't get steal time accounted, and then I truly fail to understand what you are doing. So I looked at something like this in the past. To make sure things haven't changed I set up a cgroup on my test server running a kernel built from the latest tip tree. [root]# cat cpu.cfs_quota_us 5 [root]# cat cpu.cfs_period_us 10 [root]# cat cpuset.cpus 1 [root]# cat cpuset.mems 0 Next I put the PID from the cpu thread into tasks. When I start a script that will hog the cpu I see the following in top on the guest Cpu(s): 1.9%us, 0.0%sy, 0.0%ni, 0.0%id, 0.0%wa, 48.3%hi, 0.0%si, 49.8%st So the steal time here is in line with the bandwidth control settings. Ok. So I was wrong in my hunch that it would be outside the runqueue, therefore work automatically. Still, the host kernel has all the information in cgroups. So then the steal time did not show on the guest. You have no value that needs to be passed around. What I did not like about this approach was * only works for cfs bandwidth control. If another type of hard limit was added to the kernel the code would potentially need to change. This is true for almost everything we have in the kernel! It is *very* unlikely for other bandwidth control mechanism to ever appear. If it ever does, it's *their* burden to make sure it works for steal time (provided it is merged). Code in tree gets precedence. Ok, I will work on a patch that uses the cgroup information for bandwidth control to separate out the time. * This approach doesn't help if the limits are set by overcommitting the cpus. It is my understanding that this is a common approach. I can't say anything about commonality, but common or not, it is a *crazy* approach. When you simply overcommit, you have no way to differentiate between intended steal time and non-intended steal time. Moreover, when you overcommit, your cpu usage will vary over time. If two guests use the cpu to their full power, you will have 50 % each. But if one of them slows down, the other gets more. What is your entitlement value? How do you define this? And then after you define it, you end up using more than this, what is your cpu usage? 130 %? yes exactly you would ideally show a boosted amount of cpu. However to do that you would need to either create a new tool or modify the current accounting tools such as top. My understanding is that you are not capping in this case as much as you are guaranteeing a minimum level of performance. The only sane way to do it, is to communicate this value to the kernel somehow. The bandwidth controller is the interface we have for that. So everybody that wants to *intentionally* overcommit needs to communicate this to the controller. IOW: Any sane configuration should be explicit about your capping. Add an ioctl to communicate the consign limit to the host. This definitely should go away. More specifically, *whatever* way we use to cap the processor, the host system will have all the information at all times. I'm not understanding that comment. If you are capping by simply controlling the amount of overcommit on the host then wouldn't you still need some value to indicate the desired amount. No, that is just crazy, and I don't like it a single bit. So in the light of it: Whatever capping mechanism we have, we need to be explicit about the expected entitlement. At this point, the kernel already knows what it is, and needs no extra ioctls or anything like that. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
I am deeply sorry. I was busy first time I read this, so I postponed answering and ended up forgetting. Sorry include/linux/sched.h: unsigned long long run_delay; /* time spent waiting on a runqueue */ So if you are out of the runqueue, you won't get steal time accounted, and then I truly fail to understand what you are doing. So I looked at something like this in the past. To make sure things haven't changed I set up a cgroup on my test server running a kernel built from the latest tip tree. [root]# cat cpu.cfs_quota_us 5 [root]# cat cpu.cfs_period_us 10 [root]# cat cpuset.cpus 1 [root]# cat cpuset.mems 0 Next I put the PID from the cpu thread into tasks. When I start a script that will hog the cpu I see the following in top on the guest Cpu(s): 1.9%us, 0.0%sy, 0.0%ni, 0.0%id, 0.0%wa, 48.3%hi, 0.0%si, 49.8%st So the steal time here is in line with the bandwidth control settings. Ok. So I was wrong in my hunch that it would be outside the runqueue, therefore work automatically. Still, the host kernel has all the information in cgroups. So then the steal time did not show on the guest. You have no value that needs to be passed around. What I did not like about this approach was * only works for cfs bandwidth control. If another type of hard limit was added to the kernel the code would potentially need to change. This is true for almost everything we have in the kernel! It is *very* unlikely for other bandwidth control mechanism to ever appear. If it ever does, it's *their* burden to make sure it works for steal time (provided it is merged). Code in tree gets precedence. * This approach doesn't help if the limits are set by overcommitting the cpus. It is my understanding that this is a common approach. I can't say anything about commonality, but common or not, it is a *crazy* approach. When you simply overcommit, you have no way to differentiate between intended steal time and non-intended steal time. Moreover, when you overcommit, your cpu usage will vary over time. If two guests use the cpu to their full power, you will have 50 % each. But if one of them slows down, the other gets more. What is your entitlement value? How do you define this? And then after you define it, you end up using more than this, what is your cpu usage? 130 %? The only sane way to do it, is to communicate this value to the kernel somehow. The bandwidth controller is the interface we have for that. So everybody that wants to *intentionally* overcommit needs to communicate this to the controller. IOW: Any sane configuration should be explicit about your capping. Add an ioctl to communicate the consign limit to the host. This definitely should go away. More specifically, *whatever* way we use to cap the processor, the host system will have all the information at all times. I'm not understanding that comment. If you are capping by simply controlling the amount of overcommit on the host then wouldn't you still need some value to indicate the desired amount. No, that is just crazy, and I don't like it a single bit. So in the light of it: Whatever capping mechanism we have, we need to be explicit about the expected entitlement. At this point, the kernel already knows what it is, and needs no extra ioctls or anything like that. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
On 11/28/2012 02:55 PM, Glauber Costa wrote: On 11/28/2012 10:43 PM, Michael Wolf wrote: On 11/27/2012 05:24 PM, Marcelo Tosatti wrote: On Mon, Nov 26, 2012 at 02:36:24PM -0600, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. The definition of stolen time is 'time during which the virtual CPU is runnable to not running'. Overcommit is the main scenario which steal time helps to detect. Can you describe the 'capped' case? In the capped case, the time that the guest spends waiting due to it having used its full allottment of time shows up as steal time. The way my patchset currently stands is that you would set up the bandwidth control and you would have to pass it a matching value from qemu. In the future, it would be possible to have something parse the bandwidth setting and automatically adjust the setting in the host used for steal time reporting. Ok, so correct me if I am wrong, but I believe you would be using something like the bandwidth capper in the cpu cgroup to set those entitlements, right? Yes, in the context above I'm referring to the cfs bandwidth control. Some time has passed since I last looked into it, but IIRC, after you get are out of your quota, you should be out of the runqueue. In the lovely world of KVM, we approximate steal time as runqueue time: arch/x86/kvm/x86.c: delta = current-sched_info.run_delay - vcpu-arch.st.last_steal; vcpu-arch.st.last_steal = current-sched_info.run_delay; vcpu-arch.st.accum_steal = delta; include/linux/sched.h: unsigned long long run_delay; /* time spent waiting on a runqueue */ So if you are out of the runqueue, you won't get steal time accounted, and then I truly fail to understand what you are doing. So I looked at something like this in the past. To make sure things haven't changed I set up a cgroup on my test server running a kernel built from the latest tip tree. [root]# cat cpu.cfs_quota_us 5 [root]# cat cpu.cfs_period_us 10 [root]# cat cpuset.cpus 1 [root]# cat cpuset.mems 0 Next I put the PID from the cpu thread into tasks. When I start a script that will hog the cpu I see the following in top on the guest Cpu(s): 1.9%us, 0.0%sy, 0.0%ni, 0.0%id, 0.0%wa, 48.3%hi, 0.0%si, 49.8%st So the steal time here is in line with the bandwidth control settings. In case I am wrong, and run_delay also includes the time you can't run because you are out of capacity, then maybe what we should do, is to just subtract it from run_delay in kvm/x86.c before we pass it on. In summary: About a year ago I was playing with this patch. It is out of date now but will give you an idea of what I was looking at. kernel/sched_fair.c |4 ++-- kernel/sched_stats.h |7 ++- 2 files changed, 8 insertions(+), 3 deletions(-) diff --git a/kernel/sched_fair.c b/kernel/sched_fair.c index 5c9e679..a837e4e 100644 --- a/kernel/sched_fair.c +++ b/kernel/sched_fair.c @@ -707,7 +707,7 @@ account_entity_dequeue(struct cfs_rq *cfs_rq, struct sched_entity *se) #ifdef CONFIG_FAIR_GROUP_SCHED /* we need this in update_cfs_load and load-balance functions below */ -static inline int throttled_hierarchy(struct cfs_rq *cfs_rq); +inline int throttled_hierarchy(struct cfs_rq *cfs_rq); # ifdef CONFIG_SMP static void update_cfs_rq_load_contribution(struct cfs_rq *cfs_rq, int global_update) @@ -1420,7 +1420,7 @@ static inline int cfs_rq_throttled(struct cfs_rq *cfs_rq) } /* check whether cfs_rq, or any parent, is throttled */ -static inline int throttled_hierarchy(struct cfs_rq *cfs_rq) +inline int throttled_hierarchy(struct cfs_rq *cfs_rq) { return cfs_rq-throttle_count; } diff --git a/kernel/sched_stats.h b/kernel/sched_stats.h index 87f9e36..e30ff26 100644 --- a/kernel/sched_stats.h +++ b/kernel/sched_stats.h @@ -213,14 +213,19 @@ static inline void sched_info_queued(struct task_struct *t) * sched_info_queued() to mark that it has now again started waiting on * the runqueue. */ +extern inline int throttled_hierarchy(struct cfs_rq *cfs_rq); static inline void sched_info_depart(struct task_struct *t) { +struct task_group *tg = task_group(t); +struct cfs_rq *cfs_rq; unsigned long long delta = task_rq(t)-clock - t-sched_info.last_arrival; +cfs_rq = tg-cfs_rq[smp_processor_id()]; rq_sched_info_depart(task_rq(t), delta); -if (t-state == TASK_RUNNING) + +if (t-state == TASK_RUNNING !throttled_hierarchy(cfs_rq)) sched_info_queued(t); } So then the steal time did not show on the guest. You have no value that needs to be passed around. What I did not like about this approach was * only works for cfs bandwidth control. If another type of hard limit was added to the kernel the code would potentially need to change. * This approach doesn't help if the limits are set by
Re: [PATCH 0/5] Alter steal time reporting in KVM
On 11/27/2012 07:10 PM, Michael Wolf wrote: On 11/27/2012 02:48 AM, Glauber Costa wrote: Hi, On 11/27/2012 12:36 AM, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. This can cause confusion for the end user. To ease the confusion this patch set adds the idea of consigned (expected steal) time. The host will separate the consigned time from the steal time. The consignment limit passed to the host will be the amount of steal time expected within a fixed period of time. Any other steal time accruing during that period will show as the traditional steal time. If you submit this again, please include a version number in your series. Will do. The patchset was sent twice yesterday by mistake. Got an error the first time and didn't think the patches went out. This has been corrected. It would also be helpful to include a small changelog about what changed between last version and this version, so we could focus on that. yes, will do that. When I took the RFC off the patches I was looking at it as a new patchset which was a mistake. I will make sure to add a changelog when I submit again. As for the rest, I answered your previous two submissions saying I don't agree with the concept. If you hadn't changed anything, resending it won't change my mind. I could of course, be mistaken or misguided. But I had also not seen any wave of support in favor of this previously, so basically I have no new data to make me believe I should see it any differently. Let's try this again: * Rik asked you in your last submission how does ppc handle this. You said, and I quote: In the case of lpar on POWER systems they simply report steal time and do not alter it in any way. They do however report how much processor is assigned to the partition and that information is in /proc/ppc64/lparcfg. Yes, but we still get questions from users asking what is steal time? why am I seeing this? Now, that is a *way* more sensible thing to do. Much more. Confusing users is something extremely subjective. This is specially true about concepts that are know for quite some time, like steal time. If you out of a sudden change the meaning of this, it is sure to confuse a lot more users than it would clarify. Something like this could certainly be done. But when I was submitting the patch set as an RFC then qemu was passing a cpu percentage that would be used by the guest kernel to adjust the steal time. This percentage was being stored on the guest as a sysctl value. Avi stated he didn't like that kind of coupling, and that the value could get out of sync. Anthony stated The guest shouldn't need to know it's entitlement. Or at least, it's up to a management tool to report that in a way that's meaningful for the guest. So perhaps I misunderstood what they were suggesting, but I took it to mean that they did not want the guest to know what the entitlement was. That the host should take care of it and just report the already adjusted data to the guest. So in this version of the code the host would use a set period for a timer and be passed essentially a number of ticks of expected steal time. The host would then use the timer to break out the steal time into consigned and steal buckets which would be reported to the guest. Both the consigned and the steal would be reported via /proc/stat. So anyone needing to see total time away could add the two fields together. The user, however, when using tools like top or vmstat would see the usage based on what the guest is entitled to. Do you have suggestions for how I can build consensus around one of the two approaches? Before I answer this, can you please detail which mechanism are you using to enforce the entitlement? Is it the cgroup cpu controller, or something else? -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
On 11/27/2012 05:24 PM, Marcelo Tosatti wrote: On Mon, Nov 26, 2012 at 02:36:24PM -0600, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. The definition of stolen time is 'time during which the virtual CPU is runnable to not running'. Overcommit is the main scenario which steal time helps to detect. Can you describe the 'capped' case? In the capped case, the time that the guest spends waiting due to it having used its full allottment of time shows up as steal time. The way my patchset currently stands is that you would set up the bandwidth control and you would have to pass it a matching value from qemu. In the future, it would be possible to have something parse the bandwidth setting and automatically adjust the setting in the host used for steal time reporting. This can cause confusion for the end user. To ease the confusion this patch set adds the idea of consigned (expected steal) time. The host will separate the consigned time from the steal time. The consignment limit passed to the host will be the amount of steal time expected within a fixed period of time. Any other steal time accruing during that period will show as the traditional steal time. --- Michael Wolf (5): Alter the amount of steal time reported by the guest. Expand the steal time msr to also contain the consigned time. Add the code to send the consigned time from the host to the guest Add a timer to allow the separation of consigned from steal time. Add an ioctl to communicate the consign limit to the host. arch/x86/include/asm/kvm_host.h | 11 +++ arch/x86/include/asm/kvm_para.h |3 +- arch/x86/include/asm/paravirt.h |4 +-- arch/x86/include/asm/paravirt_types.h |2 + arch/x86/kernel/kvm.c |8 ++--- arch/x86/kernel/paravirt.c|4 +-- arch/x86/kvm/x86.c| 50 - fs/proc/stat.c|9 +- include/linux/kernel_stat.h |2 + include/linux/kvm_host.h |2 + include/uapi/linux/kvm.h |2 + kernel/sched/core.c | 10 ++- kernel/sched/cputime.c| 21 +- kernel/sched/sched.h |2 + virt/kvm/kvm_main.c |7 + 15 files changed, 120 insertions(+), 17 deletions(-) -- Signature -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
On 11/28/2012 02:45 AM, Glauber Costa wrote: On 11/27/2012 07:10 PM, Michael Wolf wrote: On 11/27/2012 02:48 AM, Glauber Costa wrote: Hi, On 11/27/2012 12:36 AM, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. This can cause confusion for the end user. To ease the confusion this patch set adds the idea of consigned (expected steal) time. The host will separate the consigned time from the steal time. The consignment limit passed to the host will be the amount of steal time expected within a fixed period of time. Any other steal time accruing during that period will show as the traditional steal time. If you submit this again, please include a version number in your series. Will do. The patchset was sent twice yesterday by mistake. Got an error the first time and didn't think the patches went out. This has been corrected. It would also be helpful to include a small changelog about what changed between last version and this version, so we could focus on that. yes, will do that. When I took the RFC off the patches I was looking at it as a new patchset which was a mistake. I will make sure to add a changelog when I submit again. As for the rest, I answered your previous two submissions saying I don't agree with the concept. If you hadn't changed anything, resending it won't change my mind. I could of course, be mistaken or misguided. But I had also not seen any wave of support in favor of this previously, so basically I have no new data to make me believe I should see it any differently. Let's try this again: * Rik asked you in your last submission how does ppc handle this. You said, and I quote: In the case of lpar on POWER systems they simply report steal time and do not alter it in any way. They do however report how much processor is assigned to the partition and that information is in /proc/ppc64/lparcfg. Yes, but we still get questions from users asking what is steal time? why am I seeing this? Now, that is a *way* more sensible thing to do. Much more. Confusing users is something extremely subjective. This is specially true about concepts that are know for quite some time, like steal time. If you out of a sudden change the meaning of this, it is sure to confuse a lot more users than it would clarify. Something like this could certainly be done. But when I was submitting the patch set as an RFC then qemu was passing a cpu percentage that would be used by the guest kernel to adjust the steal time. This percentage was being stored on the guest as a sysctl value. Avi stated he didn't like that kind of coupling, and that the value could get out of sync. Anthony stated The guest shouldn't need to know it's entitlement. Or at least, it's up to a management tool to report that in a way that's meaningful for the guest. So perhaps I misunderstood what they were suggesting, but I took it to mean that they did not want the guest to know what the entitlement was. That the host should take care of it and just report the already adjusted data to the guest. So in this version of the code the host would use a set period for a timer and be passed essentially a number of ticks of expected steal time. The host would then use the timer to break out the steal time into consigned and steal buckets which would be reported to the guest. Both the consigned and the steal would be reported via /proc/stat. So anyone needing to see total time away could add the two fields together. The user, however, when using tools like top or vmstat would see the usage based on what the guest is entitled to. Do you have suggestions for how I can build consensus around one of the two approaches? Before I answer this, can you please detail which mechanism are you using to enforce the entitlement? Is it the cgroup cpu controller, or something else? It is setup using cpu overcommit. But the request was for something that would work in both the overcommit environment as well as when hard capping is being used. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
Glauber Costa glom...@parallels.com writes: Hi, On 11/27/2012 12:36 AM, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. This can cause confusion for the end user. To ease the confusion this patch set adds the idea of consigned (expected steal) time. The host will separate the consigned time from the steal time. The consignment limit passed to the host will be the amount of steal time expected within a fixed period of time. Any other steal time accruing during that period will show as the traditional steal time. If you submit this again, please include a version number in your series. It would also be helpful to include a small changelog about what changed between last version and this version, so we could focus on that. As for the rest, I answered your previous two submissions saying I don't agree with the concept. If you hadn't changed anything, resending it won't change my mind. I could of course, be mistaken or misguided. But I had also not seen any wave of support in favor of this previously, so basically I have no new data to make me believe I should see it any differently. Let's try this again: * Rik asked you in your last submission how does ppc handle this. You said, and I quote: In the case of lpar on POWER systems they simply report steal time and do not alter it in any way. They do however report how much processor is assigned to the partition and that information is in /proc/ppc64/lparcfg. This only is helpful for static entitlements. But if we allow dynamic entitlements--which is a very useful feature, think buying an online upgrade in a cloud environment--then you need to account for entitlement loss at the same place where you do the rest of the accounting: in /proc/stat. Now, that is a *way* more sensible thing to do. Much more. Confusing users is something extremely subjective. This is specially true about concepts that are know for quite some time, like steal time. If you out of a sudden change the meaning of this, it is sure to confuse a lot more users than it would clarify. I'll bring you a nice bottle of scotch at the next KVM Forum if you can find me one user that can accurately describe what steal time is. The semantics are so incredibly subtle that I have a hard time believing anyone actually understands what it means today. Regards, Anthony Liguori --- Michael Wolf (5): Alter the amount of steal time reported by the guest. Expand the steal time msr to also contain the consigned time. Add the code to send the consigned time from the host to the guest Add a timer to allow the separation of consigned from steal time. Add an ioctl to communicate the consign limit to the host. arch/x86/include/asm/kvm_host.h | 11 +++ arch/x86/include/asm/kvm_para.h |3 +- arch/x86/include/asm/paravirt.h |4 +-- arch/x86/include/asm/paravirt_types.h |2 + arch/x86/kernel/kvm.c |8 ++--- arch/x86/kernel/paravirt.c|4 +-- arch/x86/kvm/x86.c| 50 - fs/proc/stat.c|9 +- include/linux/kernel_stat.h |2 + include/linux/kvm_host.h |2 + include/uapi/linux/kvm.h |2 + kernel/sched/core.c | 10 ++- kernel/sched/cputime.c| 21 +- kernel/sched/sched.h |2 + virt/kvm/kvm_main.c |7 + 15 files changed, 120 insertions(+), 17 deletions(-) -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
On 11/28/2012 10:43 PM, Michael Wolf wrote: On 11/27/2012 05:24 PM, Marcelo Tosatti wrote: On Mon, Nov 26, 2012 at 02:36:24PM -0600, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. The definition of stolen time is 'time during which the virtual CPU is runnable to not running'. Overcommit is the main scenario which steal time helps to detect. Can you describe the 'capped' case? In the capped case, the time that the guest spends waiting due to it having used its full allottment of time shows up as steal time. The way my patchset currently stands is that you would set up the bandwidth control and you would have to pass it a matching value from qemu. In the future, it would be possible to have something parse the bandwidth setting and automatically adjust the setting in the host used for steal time reporting. Ok, so correct me if I am wrong, but I believe you would be using something like the bandwidth capper in the cpu cgroup to set those entitlements, right? Some time has passed since I last looked into it, but IIRC, after you get are out of your quota, you should be out of the runqueue. In the lovely world of KVM, we approximate steal time as runqueue time: arch/x86/kvm/x86.c: delta = current-sched_info.run_delay - vcpu-arch.st.last_steal; vcpu-arch.st.last_steal = current-sched_info.run_delay; vcpu-arch.st.accum_steal = delta; include/linux/sched.h: unsigned long long run_delay; /* time spent waiting on a runqueue */ So if you are out of the runqueue, you won't get steal time accounted, and then I truly fail to understand what you are doing. In case I am wrong, and run_delay also includes the time you can't run because you are out of capacity, then maybe what we should do, is to just subtract it from run_delay in kvm/x86.c before we pass it on. In summary: Alter the amount of steal time reported by the guest. Maybe this should go away. Expand the steal time msr to also contain the consigned time. Maybe this should go away Add the code to send the consigned time from the host to the guest This definitely should be heavily modified Add a timer to allow the separation of consigned from steal time. Maybe this should go away Add an ioctl to communicate the consign limit to the host. This definitely should go away. More specifically, *whatever* way we use to cap the processor, the host system will have all the information at all times. -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
Hi, On 11/27/2012 12:36 AM, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. This can cause confusion for the end user. To ease the confusion this patch set adds the idea of consigned (expected steal) time. The host will separate the consigned time from the steal time. The consignment limit passed to the host will be the amount of steal time expected within a fixed period of time. Any other steal time accruing during that period will show as the traditional steal time. If you submit this again, please include a version number in your series. It would also be helpful to include a small changelog about what changed between last version and this version, so we could focus on that. As for the rest, I answered your previous two submissions saying I don't agree with the concept. If you hadn't changed anything, resending it won't change my mind. I could of course, be mistaken or misguided. But I had also not seen any wave of support in favor of this previously, so basically I have no new data to make me believe I should see it any differently. Let's try this again: * Rik asked you in your last submission how does ppc handle this. You said, and I quote: In the case of lpar on POWER systems they simply report steal time and do not alter it in any way. They do however report how much processor is assigned to the partition and that information is in /proc/ppc64/lparcfg. Now, that is a *way* more sensible thing to do. Much more. Confusing users is something extremely subjective. This is specially true about concepts that are know for quite some time, like steal time. If you out of a sudden change the meaning of this, it is sure to confuse a lot more users than it would clarify. --- Michael Wolf (5): Alter the amount of steal time reported by the guest. Expand the steal time msr to also contain the consigned time. Add the code to send the consigned time from the host to the guest Add a timer to allow the separation of consigned from steal time. Add an ioctl to communicate the consign limit to the host. arch/x86/include/asm/kvm_host.h | 11 +++ arch/x86/include/asm/kvm_para.h |3 +- arch/x86/include/asm/paravirt.h |4 +-- arch/x86/include/asm/paravirt_types.h |2 + arch/x86/kernel/kvm.c |8 ++--- arch/x86/kernel/paravirt.c|4 +-- arch/x86/kvm/x86.c| 50 - fs/proc/stat.c|9 +- include/linux/kernel_stat.h |2 + include/linux/kvm_host.h |2 + include/uapi/linux/kvm.h |2 + kernel/sched/core.c | 10 ++- kernel/sched/cputime.c| 21 +- kernel/sched/sched.h |2 + virt/kvm/kvm_main.c |7 + 15 files changed, 120 insertions(+), 17 deletions(-) -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
On 11/27/2012 02:48 AM, Glauber Costa wrote: Hi, On 11/27/2012 12:36 AM, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. This can cause confusion for the end user. To ease the confusion this patch set adds the idea of consigned (expected steal) time. The host will separate the consigned time from the steal time. The consignment limit passed to the host will be the amount of steal time expected within a fixed period of time. Any other steal time accruing during that period will show as the traditional steal time. If you submit this again, please include a version number in your series. Will do. The patchset was sent twice yesterday by mistake. Got an error the first time and didn't think the patches went out. This has been corrected. It would also be helpful to include a small changelog about what changed between last version and this version, so we could focus on that. yes, will do that. When I took the RFC off the patches I was looking at it as a new patchset which was a mistake. I will make sure to add a changelog when I submit again. As for the rest, I answered your previous two submissions saying I don't agree with the concept. If you hadn't changed anything, resending it won't change my mind. I could of course, be mistaken or misguided. But I had also not seen any wave of support in favor of this previously, so basically I have no new data to make me believe I should see it any differently. Let's try this again: * Rik asked you in your last submission how does ppc handle this. You said, and I quote: In the case of lpar on POWER systems they simply report steal time and do not alter it in any way. They do however report how much processor is assigned to the partition and that information is in /proc/ppc64/lparcfg. Yes, but we still get questions from users asking what is steal time? why am I seeing this? Now, that is a *way* more sensible thing to do. Much more. Confusing users is something extremely subjective. This is specially true about concepts that are know for quite some time, like steal time. If you out of a sudden change the meaning of this, it is sure to confuse a lot more users than it would clarify. Something like this could certainly be done. But when I was submitting the patch set as an RFC then qemu was passing a cpu percentage that would be used by the guest kernel to adjust the steal time. This percentage was being stored on the guest as a sysctl value. Avi stated he didn't like that kind of coupling, and that the value could get out of sync. Anthony stated The guest shouldn't need to know it's entitlement. Or at least, it's up to a management tool to report that in a way that's meaningful for the guest. So perhaps I misunderstood what they were suggesting, but I took it to mean that they did not want the guest to know what the entitlement was. That the host should take care of it and just report the already adjusted data to the guest. So in this version of the code the host would use a set period for a timer and be passed essentially a number of ticks of expected steal time. The host would then use the timer to break out the steal time into consigned and steal buckets which would be reported to the guest. Both the consigned and the steal would be reported via /proc/stat. So anyone needing to see total time away could add the two fields together. The user, however, when using tools like top or vmstat would see the usage based on what the guest is entitled to. Do you have suggestions for how I can build consensus around one of the two approaches? --- Michael Wolf (5): Alter the amount of steal time reported by the guest. Expand the steal time msr to also contain the consigned time. Add the code to send the consigned time from the host to the guest Add a timer to allow the separation of consigned from steal time. Add an ioctl to communicate the consign limit to the host. arch/x86/include/asm/kvm_host.h | 11 +++ arch/x86/include/asm/kvm_para.h |3 +- arch/x86/include/asm/paravirt.h |4 +-- arch/x86/include/asm/paravirt_types.h |2 + arch/x86/kernel/kvm.c |8 ++--- arch/x86/kernel/paravirt.c|4 +-- arch/x86/kvm/x86.c| 50 - fs/proc/stat.c|9 +- include/linux/kernel_stat.h |2 + include/linux/kvm_host.h |2 + include/uapi/linux/kvm.h |2 + kernel/sched/core.c | 10 ++- kernel/sched/cputime.c| 21 +- kernel/sched/sched.h |2 + virt/kvm/kvm_main.c |7 + 15 files changed, 120 insertions(+), 17 deletions(-) -- To
Re: [PATCH 0/5] Alter steal time reporting in KVM
On Mon, Nov 26, 2012 at 02:36:24PM -0600, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. The definition of stolen time is 'time during which the virtual CPU is runnable to not running'. Overcommit is the main scenario which steal time helps to detect. Can you describe the 'capped' case? This can cause confusion for the end user. To ease the confusion this patch set adds the idea of consigned (expected steal) time. The host will separate the consigned time from the steal time. The consignment limit passed to the host will be the amount of steal time expected within a fixed period of time. Any other steal time accruing during that period will show as the traditional steal time. --- Michael Wolf (5): Alter the amount of steal time reported by the guest. Expand the steal time msr to also contain the consigned time. Add the code to send the consigned time from the host to the guest Add a timer to allow the separation of consigned from steal time. Add an ioctl to communicate the consign limit to the host. arch/x86/include/asm/kvm_host.h | 11 +++ arch/x86/include/asm/kvm_para.h |3 +- arch/x86/include/asm/paravirt.h |4 +-- arch/x86/include/asm/paravirt_types.h |2 + arch/x86/kernel/kvm.c |8 ++--- arch/x86/kernel/paravirt.c|4 +-- arch/x86/kvm/x86.c| 50 - fs/proc/stat.c|9 +- include/linux/kernel_stat.h |2 + include/linux/kvm_host.h |2 + include/uapi/linux/kvm.h |2 + kernel/sched/core.c | 10 ++- kernel/sched/cputime.c| 21 +- kernel/sched/sched.h |2 + virt/kvm/kvm_main.c |7 + 15 files changed, 120 insertions(+), 17 deletions(-) -- Signature -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html
Re: [PATCH 0/5] Alter steal time reporting in KVM
On Tue, Nov 27, 2012 at 09:24:42PM -0200, Marcelo Tosatti wrote: On Mon, Nov 26, 2012 at 02:36:24PM -0600, Michael Wolf wrote: In the case of where you have a system that is running in a capped or overcommitted environment the user may see steal time being reported in accounting tools such as top or vmstat. The definition of stolen time is 'time during which the virtual CPU is runnable to not running'. Overcommit is the main scenario which steal time helps to detect. Meant 'runnable but not running'. Can you describe the 'capped' case? -- To unsubscribe from this list: send the line unsubscribe kvm in the body of a message to majord...@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html