On 9/11/19 6:15 PM, Waiman Long wrote: > On 9/11/19 6:03 PM, Mike Kravetz wrote: >> On 9/11/19 8:44 AM, Waiman Long wrote: >>> On 9/11/19 4:14 PM, Matthew Wilcox wrote: >>>> On Wed, Sep 11, 2019 at 04:05:37PM +0100, Waiman Long wrote: >>>>> When allocating a large amount of static hugepages (~500-1500GB) on a >>>>> system with large number of CPUs (4, 8 or even 16 sockets), performance >>>>> degradation (random multi-second delays) was observed when thousands >>>>> of processes are trying to fault in the data into the huge pages. The >>>>> likelihood of the delay increases with the number of sockets and hence >>>>> the CPUs a system has. This only happens in the initial setup phase >>>>> and will be gone after all the necessary data are faulted in. >>>> Can;t the application just specify MAP_POPULATE? >>> Originally, I thought that this happened in the startup phase when the >>> pages were faulted in. The problem persists after steady state had been >>> reached though. Every time you have a new user process created, it will >>> have its own page table. >> This is still at fault time. Although, for the particular application it >> may be after the 'startup phase'. >> >>> It is the sharing of the of huge page shared >>> memory that is causing problem. Of course, it depends on how the >>> application is written. >> It may be the case that some applications would find the delays acceptable >> for the benefit of shared pmds once they reach steady state. As you say, of >> course this depends on how the application is written. >> >> I know that Oracle DB would not like it if PMD sharing is disabled for them. >> Based on what I know of their model, all processes which share PMDs perform >> faults (write or read) during the startup phase. This is in environments as >> big or bigger than you describe above. I have never looked at/for delays in >> these environments around pmd sharing (page faults), but that does not mean >> they do not exist. I will try to get the DB group to give me access to one >> of their large environments for analysis. >> >> We may want to consider making the timeout value and disable threshold user >> configurable. > Making it configurable is certainly doable. They can be sysctl > parameters so that the users can reenable PMD sharing by making those > parameters larger. I suspect that the customer's application may be generating a new process with its own address space for each transaction. That will be causing a lot of PMD sharing operations when hundreds of threads are pounding it simultaneously. I had inserted some instrumentation code to a test kernel that the customers used for testing, the number of timeouts after a certain time went up more than 20k. On the other hands, if the application is structured in such a way that there is limited number of separate address spaces with worker threads processing the transaction, PMD sharing will be less of a problem. It will be hard to convince users to make such a structural changes to their application. Cheers, Longman
