Re: What causes VM to lock pages into memory?
While there is a counter 'LOCKED' that refers to explicit locking of pages via the CP lock command. The counter I referred in a problem record Steve is describing is overall locked pages which may include implicit locking for things like QDIO data structures, traditional I/O buffers, and other CP interactions. While 2000-4000 pages may not be unusual for a Linux guest with dedicated OSA devices, the numbers in this case are larger. We recommended looking at monitor data and a snampdump at next occurence to see if we could correlate the increase in locked pages. From the data I looked at these pages were not associated with any particular guests. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Do 64-bit Linux guests eliminate VM <2GB swapping?
Sorry, should have seen this discussion earlier. Since the majority of z/VM's control program (CP) is still 31-bit addressing, a guest page that CP is working with will tend to be brought below the 2GB bar. There are always exceptions of course. See http://www.vm.ibm.com/perf/tips/2gstorag.html for additional information. So net is - 64-bit guests will not eliminate need to bring pages below 2GB (VM doesn't use 'swap' terminology). In fact there are scenarios where larger virtual machines can make it worse. The more pages a linux guest has the larger the possible number of pages that CP *may* need to process. It depends. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Do 64-bit Linux guests eliminate VM <2GB swapping?
There is no significant change in z/VM 5.1.0 that addresses this particular situation, but I should update the web page I pointed to earlier to reflect that. Les posted the SOD, which is about all I should say. If there are additional changes in the service stream, I'll do my best to let you know on this listserv. As to the question of 'what it depends on?'. :-) I hope the web page mentioned earlier answers most of that. Bit -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: linux swap to dcss?
A few thoughts on various statements: 1.The Linux development process is a little different than the traditional S390 where we test things ad nauseum before we let them out the door. I apologize that VM Performance hasn't had time to fully measure the swap to dcss. It's on the list, but there are a few more pressing issues right now. But this discussion is good; it helps me think about what to consider in the measurements. 2.A DCSS segment defined as EW will have a lower priority in the VM storage management steal algorithms, unlike traditional segments and vdisks. So it doesn't have the same challenge as vdisk currently does. 3.Swapping to DCSS can be done as a block device which can lower the pathlength (i.e. not having to create CCWs etc.) compared to vdisk swapping. 4.Using recent VM and Linux you can create gaps for the DCSS, but the DCSS is still limited below 2GB. So it won't be an acceptable alternative in all cases. 5.The swapping itself can occur without VM intervention, except for the creation and when required the actual paging of the DCSS (that's VM paging, not Linux). Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux Slowdown
A few comments on QUICKDSP; some opinion, some fact. I tend to look at QUICKDSP in three lights: 1. As Barton mentioned any guest/server that another guest depends on. He once used the phrase 'anything that is an extension of the operating system (VM)' should have quickdsp. Network, security, file systems, etc. 2. Production guests or mission critical, i.e. any guest that if it were to stop you could get a phone call from someone with authority to fire you. "Production" gets to be gray because some 'development' guests consider themselves production because they produce code. :-) 3. The more predictable or reliable a guest is, the more likely I am to give it QUICKDSP. Use of QUICKDSP doesn't really 'disable' the scheduler, it just tells the scheduler to behave differently in one aspect. When you use QUICKDSP you are saying, I (the sys programmer) am taking responsibility that this guest will not over-burden the available real resources (i.e. avoid thrashing). Test systems, by their nature, are not predictable so I tend not to give them QUICKDSP. So the two approaches (three when you do it in combination) are 1) use of QUICKDSP and 2) adjust STORBUF settings. I usually recommend considering both (after first revisiting the size of the guests as Adam recommended). When adjusting STORBUF there are some 'rules of thumb' floating around out there, but it is worthwhile understanding what the numbers mean. See the documentation or Malcolm's write up: http://www.marist.edu/htbin/wlvtype?LINUX-VM.30359 I do not generally recommend setting all three values in STORBUF the same, especially if you have a mixed environment with interactive CMS guests. It may be perfectly acceptable in pure Linux guest environment. Also, just to clarify. QUICKDSP does not disable the scheduler's ability to effectively set priorities based on Share settings. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Domino on LINUX/VM
Doug, The primary concern for running Domino in a 1-way configuraiton is to avoid a background or mgmt Domino task from taking over the system. The thought being that a second processor allows one to allow other work to run. Depending on the heaviness of your Domino users, you maybe be able to run a fair amount on a single IFL. However, you'll need to be aware of the potential to lock out end users for periods of time due to these background tasks. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Domino on LINUX/VM
Doug, that's an interesting idea to run a virtual 2-way on the single IFL. I never thought of that. It is, as you say, against conventional wisdom, but this job is full of exceptions. I could see some challenges in terms of lock contention, though this might be a case where Linux use of Diagnose x'44' actually is of benefit. I expect the trade off is slightly worse performance for all commands all the time, but a new safety net to avoid lock outs from long running tasks. I would be very interested in hearing what you see. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: RVA,ESS,performance
>You can have only one I/O in progress to each physical device at a time >(since Linux doesn't do PAV yet). If you use a logical volume comprised >of more physical devices, more actual I/Os can be in progress at the >same time. If one runs under VM, you can use PAV with Linux. See http://www10.software.ibm.com/developerworks/opensource/linux390/perf/tuning_rec_dasd_PAV.shtml#begin for details. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: RVA,ESS,performance
>If one runs under VM, you can use PAV with Linux. See Of course I should have included, this applies only to devices that support PAV, which I do not believe the RVA does. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: DASD I/O
See http://awlinux1.alphaworks.ibm.com/developerworks/linux390/perf/index.shtml (note this is a temporary link. I have http://www.ibm.com/developerworks/linux/ bookmarked.). Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Pros and cons - emulated FBA or FCP-attached SCSI???
I will try to avoid adding confusion. From a pure pathlength point of view I suspect that the FCP attached SCSI EVMS would be the better performer. But let me tack on an "It depends". :-) >From a performance analysis and management view, the traditional monitoring capabilities are much stronger for Ficon/3390 than for FCP SCSI. Both the Emulated FBA and traditional 3390s are eligible for minidisk cache (MDC). There are also a number of non-performance related pros and cons also, in terms of how you are doing backups/restores, management, cloning, etc.. And the sky here is currently endicott gray, which is a very bright and happy shad of gray. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: relative share and quick dispatch
There is a fair amount of confusion, and some disagreement in this area. I would not just add QUICKDSP to all my guests. However, there are three types of guests that should have QUICKDSP set ON: 1. Any server virtual machine that is key to operation of the system (Barton describes this as 'an extension of the operating system'). This includes your security manager, accounting, file servers, etc. 2. Critical guests- My definition: any guest that if it were to stop running for a period of time could result in someone being fired. 3. The VM userid you use to do tuning (so if you make a mistake you can undo it). The QUICKDSP setting exempts the virtual machine from having to sit time out in the eligible list. It does not exempt the virtual machine from being dropped from the dispatch list. Relative share and absolute share usage can also be a religious discussion. Relative is convenient because you can use it to give a 'relatively' higher priority without needing to necessarily understand actual resources consumed (though that helps). That is Relative 2000 virtual machine would have 20 times better access to resources than default Relative 100 share virtual machine. Now, using Absolute share requires a bit more planning upfront, but it can be necessary for certain guests. Remember Relative share is relative to which virtual machines are competing for resources. So with Relative share, as the number of competing virtual machines increases, the effective priority of a relative share virtual machine decreases. This may not be what you want. A little more discussion on this occurs at http://www.vm.ibm.com/perf/tips/linuxper.html with some links to more details. That covers the major options that can be set in the VM directory. The other command (sorry can't be set in directory) that you might want to be familiar with is CP SET RESERVE to give some favoritism of memory for a guest. I haven't seen it needed in many Linux z/VM environments, but it is worth knowing about it. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux under VM under VM (or above, if you like :-)
Please note that the more layers of virtualization you add, the greater the overhead for running the Linux guest. This includes the LPAR layer. If you are running VM on VM on LPAR, it can get very expensive. Bill Bitner - VM Performance Evaluation - IBM Endicott
Re: CPU Scalability In single Linux Image Under VM?
It depends. With Linux it depends even more. Remember that scaling involves both a software and a hardware MP factor. On zSeries, we have some advantages from the hardware implementation. There are some workloads that scale very well on Linux for zSeries. Klaus Bergmann did a presentation at the last SHARE that gave some examples of excellent scaling on a 16-way. Of course I'm sure that's not the case for all workloads. :-) As for scaling with virtual MP support. There is a small bump/cost in VM overhead in going from a virtual 1-way to a virtual n-way. However, after that the overhead is basically linear up to the maximum of 64 for a given virtual machine (though defining more virtual processors than real processors will not be of any benefit and may just add overhead). Bill Bitner - VM Performance Evaluation - IBM Endicott 607-752-6022 Internet: [EMAIL PROTECTED] Home Page: http://www.vm.ibm.com/devpages/bitner/
Re: DASD Performance problem
Are you collecting any performance data? As Barton mentioned, this workload could be taxing the processor, or you could also be gated by the single disk. Perhaps using LVM across multiple disks would help. Understanding whether this is channel, device, or processor limited would help. I assume the Dell system is also use ext3. If not that would make a difference. Also, the comparison is a little different because of using Ficon instead of FCP SCSI. However, all in all, I have seen other measurements with similar results, particularly for a single guest with a single disk. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286
need a how to...
You could also try the CP INDICATE QUEUE EXP command. One of the common gotchas is the formation of an eligible list. The CP command above will give you something like: TCPIP Q0 PS 6196/5746 -199.9 A02 BITNERQ1 R00 0819/0795 .I.. -179.9 A02 KREINTA Q3 IO 2629/3353 -.0721 A00 ABBEY Q3 R01 00022163/00022162 .0199 A01 CORAK2Q3 PS 00075799/00075799 ..D. 9 A02 CORAK2 MP07 Q3 PS / 9 A02 EDLLNX1 Q3 PS 6052/6052 9 A02 VMLINUX1 Q3 PS 1258/1258 9 A02 EDLLNX2 Q3 PS 00010855/00010855 ..D. 9 A00 DCEPKBLD Q3 PS 00019142/00019142 9 A02 SSLSERV Q3 PS 3502/3478 9 A02 VMLINUX Q3 PS 4832/4832 ..D. 9 A00 If you were to see an "E3" instead of "Q3" that would be an indication of the virtual machine being held in the eligible list because the scheduler feels that to run the virtual machine, it would overload system resources. See also http://www.vm.ibm.com/perf/tips/linuxper.html for some discussion on this and links to other info. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286
Re: VM SET REORDER OFF for large hosts
You'll find additional data on Reorder Processing at http://www.vm.ibm.com/perf/tips/reorder.html Feel free to contact me if you have other questions or want to discuss a particular scenario. Bill Bitner - z/VM Customer Focus and Care - IBM Endicott - 607-429-3286 The postings on this site are my own and don't necessarily represent IBMs positions, strategies or opinions. -- 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: Performance Toolkit and zLinux
The RMF agent that runs inside the Linux on System z guest was re-written the other year and changed the interface. We chose not to update the Performance Toolkit handling of that interface as other things have changed since the introduction of the RMF agent to Linux. You can still use the RMF client (the link provided earlier will point to information ont that) that supports z/OS as well. There was a statement of direction in 2011 for these changes. Performance Toolkit will report on the z/VM Appldata that Linux provides on all the current distributions. While this doesn't provide process level information, it does provide a Linux view of some important metrics. It is also very low overhead. Additionally, OMEGAMON XE for z/VM and Linux provides both the z/VM and the Linux agents. Perhaps not the first choice for a PoC, but wanted to mention for completeness. Bill Bitner - z/VM Customer Focus and Care - IBM Endicott - 607-429-3286 The postings on this site are my own and don't necessarily represent IBMs positions, strategies or opinions. -- 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: Automated performance reporting on Linux on Z
The IBM OMEGAMON XE on z/VM and Linux tool collects performance metrics and allows customers to store them in a data warehouse (included). Customers choose what to keep and they summarize and prune the data based on the reporting they would like to consider. Another included feature of any OMEGAMON or IBM Tivoli Monitoring component is the Cognos based reporting tool Tivoli Common Reporter (TCR) which enables custom and dynamic reporting against the data in the data warehouse. Many customers use this to produce the kind of reports you are mentioning. Not only can you report on z/VM and Linux data, but any other data in the data warehouse. If you collect middleware or database data on those Linux guests, these attributes can be included in this reporting structure. Finally, another included feature that uses this same warehouse data is the Tivoli Performance Analyzer which adds vendor supplied predictive capability focusing on future performance issues. Customers can change values and play "what if" type of scenarios. All in all, you get real-time performance analysis, custom reporting, and predictive analysis in one tool. Bill Bitner -- 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: Why zLinux, Cross posted to MVS-OE
A paper I've recommended particularly to address why System z virtualization vs. other platforms is: http://public.dhe.ibm.com/common/ssi/ecm/en/zsl03192usen/ZSL03192USEN.PDF Bill Bitner - z/VM Customer Focus and Care - IBM Endicott - 607-429-3286 The postings on this site are my own and don't necessarily represent IBMs positions, strategies or opinions. -- 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/
z/VM & Linux Support of the IBM z13
Hello Everyone, just a quick reminder that there are required service levels for both z/VM and Linux for the new IBM z13. Besides the PSP Bucket, additional information is available for z/VM at: http://www.vm.ibm.com/service/vmreqz13.html Note the above list includes support for the Multi-VSwitch Aggregation (ability to do LAG with shared OSAs) which was announced to be available June 26th, so you won't find closed PTFs at this time. The absence of the four related PTFs will not limit use of the z13, just the new VSwitch support. Linux tested environments can be found at: http://www-03.ibm.com/systems/z/os/linux/resources/testedplatforms.html Unlike some of the past machines where you could put off getting to the required levels of Linux when running as a guest, do not make that assumption this time. You might find yourself with a Linux that won't boot any more. Also please take a minute to read the footnotes on that page they are technical in nature, not just legal :). Even if you somehow managed to bring up one of the unsupported Linux levels on your new z13, you might only be temporarily lucky. Additional updates to millicode could find you in an unfortunate situation where those unsupported Linux guests no longer run. Keep Calm and Stay Current Bill Bitner -- 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/
IBM z Systms Java Survey
Posting for Marcel Mitran: Please take a few minutes to fill in the following Java on z survey, and feel free to forward to others who may be interested, or have insight. Link: https://ibm.biz/java_on_z Survey: z Systems software values your feedback! Audience: Anyone who has been interested in using Java on z Systems Purpose: Learn about our client's experience with Java and new programming languages on z Deadline: Saturday, Sept 25 Duration: 5-7 minutes Incentives: Help define the future of development on z, sign up for early program feedback Thanks, Bill -- 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: zLinux operating as a zVM guest host
To echo Mark's and others' comments. This configuration is likely to have a lot of overhead with z/VM on top of z/VM on top of LPAR. Most customers I've worked with placed your three z/VM systems in different LPARs. Partitions can be capped similar to LIMITHARD with Relative Share setting. To avoid confusion, the changes with VM64721 for the SET SHARE command do not affect Limitsoft (and were not inteded to do so). However, VM64721 does affect Limithard, but only when turned on via SET SRM LIMITHARD CONSUMPTION command. (SET SRM LIMITHARD DEADLINE is the default) Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- 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
Re: FCXPER315A message
That may not be anything to worry about. Here's the explanation from VM Perf FAQS http://www.vm.ibm.com/perf/tips/prgcom.html Problem: Performance Toolkit (or insert favorite monitor here) is giving me alerts about the C1ETS being too high. Solution: The C1ETS stands for class 1 Elapsed Time Slice. Each scheduler class has an Elapsed Time Slice (ETS) associated with it. The Class 1 ETS is dynamically adjusted by the scheduler. All the other time slices are multiples of the C1ETS (classes 0/2/3 multiplication factors are 6/8/48 respectively). The scheduler adjusts C1ETS in order to try and keep 85% of the transactions as trivial (that is within the first ETS). On systems where there are guests that never go truly idle, the transactions are very infrequent and therefore can cause the scheduler to increase the C1ETS. This isn't necessarily a problem since the transactions are not real transactions. Regards, Bill Bitner -- 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: z/VM 5.4 vs z/VM 6.1
The z/VM 5.4 and z/VM 6.1 are starting to diverge. Along with the zBX and URM aspects. We're starting to see more service of the 'enhancement' nature going into 6.1 and not 5.4. For example, an improvement to scheduler lock management VM64927 is available for 6.1, but not for 5.4. This is interesting for systems on larger n-ways that want to reduce z/VM and LPAR overhead. Bill Bitner -- 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: Expected Network Throughput from Layer-2 VSwitch Attached zLinux Instance
Hello Aaron, For guest to external network, it will be very much gated by the OSA model/speed. There are some examples for 10GbE OSA Express 3 in http://www.vm.ibm.com/devpages/bitner/presentations/vmup2011.pdf For guest to guest, it'll be a factor of config and which machines. I think we saw something around a peak of 900 MB/Sec for streaming workloads with multiple sessions; 600 MB/Sec for streaming with single session. That was with jumbo frames on MTU. Bill Bitner IBM z/VM Customer Focus and Care -- 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: Set Share Relative
IBM has talked about the excess Share problem as well, see chart 23 in http://www.vm.ibm.com/devpages/bitner/presentations/vmup2011.pdf And we are working on it. I think an assertion that it happens all the time is a little overboard. Now that we understand the problem better, I can say everyone may be exposed, but not everyone will be affected. We have been asking for people to contact me or open a PMR if they see the problem and have data. No PMRs have been open yet. A PMR always makes it easier to prioritize. I don't understand why those that see it, refuse to open a PMR. The idea that IBM is working on changing this behavior to get around too high default share settings, while creative, is fiction. I agree some share default settings are now too high for Linux environment workloads. In general, the experience has been that it is better for some of these SVMs to be too high rather than too low. I'll put it on the list to revist some of these and see if we can change where appropriate. The defaults are meant as starting points. As discussed already, there are some good papers out there to assist with the tweaking. And for some other products, such as VTAM, the defaults are wrong for 99% of the environments. Sorry, my failure to get changes made there. Bill Bitner - z/VM Customer Focus and Care - IBM Endicott - 607-429-3286 -- 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: PAV usage on DASD volumes
Hello Barry if you are using traditional PAV, you should not see any activity in Domain 6 Record 28. That record only includes activity associated with HyperPAV. Fields of interest to traditional PAV would be Domain 6 Record 3, such as: IODDEV_PAVINELIG IODDEV_PAVUSES IODDEV_RDEVPVFG Regards, Bill -- 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: zLinux performance collection tool(s)?
OMEGAMON XE has a client for Linux which is included in the OMEGAMON XE for z/VM and Linux product, or can be purchased separately. It includes process level data. The CPU Timer based accounting is an improvement regardless of whether you use normalization or not. The old method not only had the potential for inflated values, but also was exposed to skews in different directions. Normalizing the old values was a clever way of dealing with the problem in the interim, but it did not solve all the problems. Hence, the need for the CPU timer approach. Thank you Martin. :-) It is true that looking at both the VM and Linux data is helpful. Two heads are better than one. Kevin, I believe you have the VM part working already. There's been discussion here about various VM based solutions for looking inside the guest. For your needs, particularly in short term, a lot of the Linux tools out there will work on z, and with the new CPU timer are accurate. Stay away from things like TOP that are noisy and just add overhead. I should also mention that I don't see Performance Toolkit's requirement to indicate which virtual machines you want detailed data on for historic reasons as a hinderance to the solution. Inconvenient perhaps, but not a show stopper. It might be the confusion that in the old RTM product you could only benchmark one virtual machine, where Performance Toolkit allows you to benchmark any number. Also, I believe with different uses of OMEGAMON XE you can set up historical collection as well. Unfortunately I don't know the details. I hope that helps. Bill Bitner -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: SLES9 SP3 / zVM 530 Vnic capacity?
One should not expect Link Aggregation to improve the performance of a single Vnic talking to a single target. The packets associated with a single conversation (probably not the right term) are not shuffled across the different OSAs or spread across them. We do spread different conversations across different OSAs with load balancing. The z/VM Performance Report has additional details: http://www.vm.ibm.com/perf/reports/zvm/html/linkagg.html Regards, Bill Bitner -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Is 275GB of VDISK stupid?
This is a good approach, but please one reminder. The DAT structures for virtual disk in storage address spaces, including PGMBKs, are not pageable (user space PGMBKs are). And prior to z/VM 5.3.0, they must reside in real memory below 2GB. So 275GB of vdisk would be more than 2GB of memory required below 2GB for the DAT structures. One would want to be on z/VM 5.3.0. The PGMBKs are not reflected in the PTRM numbers. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Is 275GB of VDISK stupid?
>> This is a good approach, but please one reminder. The DAT structures >> for virtual disk in storage address spaces, including PGMBKs, are not >> pageable (user space PGMBKs are). And prior to z/VM 5.3.0, they must >> reside in real memory below 2GB. So 275GB of vdisk would be more than >> 2GB of memory required below 2GB for the DAT structures. One would >> want to be on z/VM 5.3.0. The PGMBKs are not reflected in the PTRM >> numbers. > >But only when used, isn't it? >As long as you have not touched them, there's no PGMBKs except for the >first one. Once he actually starts using them I would think that 550 >GB of paging space might approach some other limitations in CP (though >with 3390-9 not the number of cpowned volumes). It's been a long time since I looked at the code, so I am not certain of all the cases that trigger creation of the DAT structures. I do recall that there were cases where they could be created without the actual virtual disk block being referenced. Like I said, it's not a bad idea, I just don't want anyone who's not on 5.3.0 to be caught by surprise. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Bogus CPU Utilization Numbers reported in the PTK for Linux guest
Mark posted a nice summary (pun intended). Just one clarification, OMEAGMON XE did ship support to properly normalize the CPU values for older releases of Linux. This shipped in September 2007 as VM64299 available for both the z/VM 5.2 and 5.3 versions of Performance Toolkit. PTF UM32169 for z/VM 5.2 provides the base function for the formatted output collectors and provides an update for normalizing Linux data in the formatted output. PTF UM32170 for z/VM 5.3 provides the update for normalizing Linux data and provides support for mixed processor types in the formatted output. Also, while there are methods to normalize the old accounting information, it is still subject to some misleading in that the old method is exposed to skewing of time towards buckets. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Linux VM intermittently goes non-responsive
> IBM, have you thought about changing that default? > I bet everyone hits it eventually.=20 Hello all, yes we have. Actually I was discussing this topic last week with part of the team. First my apologies that the SRM defaults cause the problems they do. We have changed the defaults in the past, the last time being VM/ESA 1.2.2, so we're probably due. There are two challenges here. The first challenge is that there are environments where the current defaults are appropriate. I had been thinking that the correct approach was to change so that the SRM values could be set in the SYSTEM CONFIG file. We would then ship the default SYSTEM CONFIG file with SRM values that are appropriate with the idea that most existing customers would continue to use their own SYSTEM CONFIG files. Even better, down the road we could have different SYSTEM CONFIGs for different usage. The line item to add SRM values to SYSTEM CONFIG has missed the cut to be included the past couple releases. That was part of the discussion last week. We talked about just changing the defaults and forgetting the SYSTEM CONFIG file. (Though I still like the idea of having that and other things in SYSTEM CONFIG instead of hidden in different places on every system (OPERATOR PROFILE, AUTOLOG profiles, Directory entries, etc.). The second challenge is while everyone agrees they should be changed, it is far more difficult to agree on what the new values should be. I conveniently have a meeting with the scheduler team today. This is on my agenda. So again, I apologize if it appears we have been ignoring this aspect of the system. I'm probably as much to blame as anyone. I'll try to make it right. Regards, Bill Bitner -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Latest on PAV's?
Not that I want to bring this subject back up, but I think there were a few unanswered questions. You can find the device queueing on Performance Toolkit FCX108 DEVICE report under the "Req. Qued" field. Also is reflected in difference between "Serv" & "Resp" (Service Time and Response Time). Shown in example below where MDC is not enabled. <-- Device Descr. --> Mdisk Pa- <-Rate/s-> <--- Time (msec) ---> Req. Addr Type Label/ID Links ths I/O Avoid Pend Disc Conn Serv Resp CUWt Qued >> All DASD << .5.0 .2 .1 2.0 2.3 2.3 .0 .00 80C9 3390-3 US7E42 158 4 358.0 .2 .0 2.0 2.2 6.0 .0 1.37 For those still running VMPRF, it would be on the DASD_BY_ACTIVITY Report. In regards to paging to 3390-9s, yes, it would be a horrible idea if it were a real 3390-9. Otherwise, it can be done depending on the rest of the configuration. I would avoid it, if I could. But if you are approaching the max of 256 CP owned volumes: 1) I'd like to hear from you. 2) Your access density for paging is probably low enough that it doesn't matter at this point. For those wanting to read more about Linux and PAV on VM see: http://awlinux1.alphaworks.ibm.com/developerworks/linux390/perf/tuning_rec_dasd_PAV.shtml#begin In regards, to the skew of access density. Traditionally, minidisk cache addresses that and is why it was designed the way it was versus a private cache structure. But I don't agree that it is queuing on the device level that is the bulk of the I/O performance problems. The last 5 I/O problems I looked at were: 1 - channel contention on ESCON channels 2 - not balancing I/O across ranks 3 - not using minidisk cache 4 - PPRC config related 5 - application boo boo For those looking to blame someone for the lack of z/VM PAV support for minidisks and CP volumes, feel free to blame me; but be nice, my Mom & my manager might be reading this. For those that want to discuss their systems and choices, feel free to contact me. I'll respond as time permits. :-) Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: zVM 5.1, performace toolkit and Linux
On most current distributions I've looked at there is no need for a recompile. There is a Redbook, currently still in draft, that does a good job of describing the monitor data interface. See Chapter 12 of http://www.redbooks.ibm.com/redpieces/abstracts/sg246695.html?Open Performance Toolkit supports both the RMFPMS interface and the new kernel appldata interface. I don't recall seeing a requirement for the SNMP. But if folks are interested, let us know. If there is enough interest, we would consider it. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286VP -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: z/VM Minidisk Cache. Still relevant today?
MDC is a write-through cache so it does not apply to write I/Os It can help read I/Os, but the answer to how/if to use it is "it depends". As Carsten mentioned, it can be great for shared minidisks with significant read I/O. There are also cases where for non-shared minidisks, I might want to use it. While Linux does a lot of caching, that requires memory. In order to get very high hit rates at the Linux level, you need to supply enough virtual memory for its cache. The downside is Linux will use it, and increase pressure on overall system memory requirements. So there are cases where it is better to have a lower virtual machine sizes for the Linux guests and have MDC pick up some of the missed I/Os. This is a trade-off between memory (linux cache) and processor resoures (MDC handling the I/O). Another case where MDC may be helpful is that it may avoid having to bring a guest page below 2GB for processing if its an MDC hit. This can be helpful if you are constrained by memory below 2GB. When used for this approach, we sometimes suggest forcing MDC to be used from real storage, not expanded storage. I do recommend turning MDC off for write-mostly disks (log files), most Database applications, and back-up processing. Though, there are always exceptions. Other information may be found at http://www.vm.ibm.com/perf/tips/prgmdcar.html Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: Hipersockets vs qdio
In most cases, a Guest LAN or Virtual Switch Guest LAN would be the way to go. HiperSockets would be my second choice. Shared OSA third. That is for inter Guest communication. HiperSockets may be faster for very large data streaming applications, but may not be noticeably different for a lot of workloads. HiperSockets and Shared OSA would also require more locked memory than Guest LAN or Virtual Switch. See http://www.vm.ibm.com/perf/reports/zvm/html/ for detailed measurements. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: What configurations are people using for Disaster Recovery for Linux under z/VM
I do not believe %SIC means what you think it means. I had a conversation about this in the past I think with someone from the County if it is the same configuration. At the DR location, if you are running Linux as a guest of VM and VM running as a guest of another VM system and all that running in an LPAR, then you effectively have 3 layers of SIE. (1 for each VM and 1 for LPAR). Current machines can handle 2 layers of SIE very efficiently, but performance does drop off significantly with 3 layers of SIE. SIE is how VM (and LPAR) dispatches work. We sometimes use the phrase SIE Break or Exit from SIE to describe the cases where the virtual (or logical) processor stops running under SIE and control is passed back to the hipervisor. There are two reasons to exit SIE: intercepts and interrupts. The %SIC is what percentage of the exits were for intercepts. So the number could be high here because of exits for SIEs done under the three levels, but it could also be influenced by CPU bound work or work that naturally has low level of interrupts. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: MDISK vs DEDICATED DASD
Dedicated DASD are not eligible for VM's minidisk cache. -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: FCP over ECKD performance advantage - why?
It depends. From a response time perspective, the bulk of the difference is dependent on the hardware. FICON/FCP to DS8000 both will likely give similar response time. Processor time, particularly in a virtualized environment will vary significantly. See http://www.vm.ibm.com/perf/reports/zvm/html/520lxd.html In general on system z, FCP has three areas of potential advantage: 1. avoids overhead of converting from block to eckd and back to block oriented in the CU. 2. more I/Os can be executed in parallel (though PAV is a method for ECKD to minimize this advantage) 3. more data can be moved in a single I/O command Workloads are impacted differently by the above. There are also a number of non-performance related differences. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: collaborative memory management on Linux weekly news
When the VM support is released there will be additional monitor counters; and we do plan on publishing additional results (in addition to what was published on the referenced article and at various conferences). Part of the benefit is not just saved memory, but more efficient management of that memory. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: FCP over ECKD performance advantage - why?
Rob is correct it does say 'it depends', but gives one of the dependencies - processor time. The report also mentions that there are other considerations (trade-offs) such as performance information being reported (as Barton mentioned). Also, you need to consider: backup/recovery, management/provisioning, disaster recovery, and data administration. Many of these considerations are not performance related, the report only tries to address performance, and even one aspect of that in detail. But I hope it is helpful. Bill -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
Re: SET SHARE recommendations
Hello Kevin, First, I would question do you really need 5 virtual processors for each of these guests? You might, but I know some people just automatically set the virtual processor count equal to the logical processor count and in cases that is not optimal. Ok, I'll resist giving exact numbers, but a few thoughts: 1. Remember the share setting is distributed across the virtual processors. So a virtual 5-way with default relative 100 gets dispatched basically as five relative 20 virtual processors. 2. If you are going to have this finite number of guests, relative shares might be good to use. (The more 'extra' guests you get, the more inclined I would be to use absolute shares for the production guests and relative for the cert guests.) 3. For the Production guests, I would use Relative shares for target minimum share values and No limit shares. 4. For the Cert guests I would set fairly low Relative shares for the target minimums and use LIMITSOFT settings for limit share. Bill Bitner - VM Performance Evaluation - IBM Endicott - 607-429-3286 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
z/VM 6.3 End of Service Reminder
z/VM 6.3 will achieve end of service on December 31, 2017, leaving only z/VM 6.4 in service. (z/VM 6.2 is EOS on June 30, 2017.) For full information on life cycle see: https://www-01.ibm.com/software/support/lifecycle/ For clients who missed this or clients looking for guidance on migration to z/VM 6.4, you can contact Bill Bitner (bitn...@us.ibm.com). Various options exist: Extended support contracts for z/VM 6.3 IBM Lab Services for additional temporary skills Also remember that z/VM 6.4 introduced a new architecture level set, requiring a z196 or z114 or higher machine. Tell a friend. Have a nice weekend. Bill Bitner -- 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: High steal on z/linux guest
Take a look at the 4th entry on this page - http://www.vm.ibm.com/perf/tips/prgcom.html Regards, Bill Bitner -- 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: Meltdown/Spectre; Linux on z affected?
Hello folks, sorry, but you know the drill. This kind of information is shared only through the IBM Z Security Portal. If you're unfamiliar with it or how to sign up, this was covered in part of the Live Virtual Class on October 18, 2017 that Brian Hugenbruch did. http://www.vm.ibm.com/education/lvc/zvmlvc.html Thanks and have a safe day. Regards, Bill -- 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: Guest show steal and the LPAR has available CPU
See the fourth item on this page: http://www.vm.ibm.com/perf/tips/prgcom.html ___ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- 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: Guest show steal and the LPAR has available CPU
I believe in Velocity it is the ESAXACT report. ___ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From: Victor Echavarry To: LINUX-390@VM.MARIST.EDU Date: 02/01/2018 04:27 PM Subject:Re: Guest show steal and the LPAR has available CPU Sent by:Linux on 390 Port Bill: Which report is z/VM state sampling? Right now we are using velocity to monitoring our z/VM's. Regards, Victor Echavarry System Programmer Operating Systems EVERTEC, LLC -Original Message- From: Linux on 390 Port [mailto:LINUX-390@VM.MARIST.EDU] On Behalf Of Bill Bitner Sent: Thursday, February 1, 2018 3:58 PM To: LINUX-390@VM.MARIST.EDU Subject: Re: Guest show steal and the LPAR has available CPU See the fourth item on this page: http://www.vm.ibm.com/perf/tips/prgcom.html _______ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit https://urldefense.proofpoint.com/v2/url?u=http-3A__www.marist.edu_htbin_wlvindex-3FLINUX-2D390&d=DwIFAg&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=zIUVOGBLCjZgWnU3DmF7lZT7l1bWZgu2QxvNY7dvZdo&s=aBMZ09Pqgp_6XxwrnmjegxtTbjtgS4qyzCWtgPJFh60&e= -- For more information on Linux on System z, visit https://urldefense.proofpoint.com/v2/url?u=http-3A__wiki.linuxvm.org_&d=DwIFAg&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=zIUVOGBLCjZgWnU3DmF7lZT7l1bWZgu2QxvNY7dvZdo&s=PwLIdR5U1lP1jLVBdj1mcn1CUxZcCn7tkQooDnMokUQ&e= -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit https://urldefense.proofpoint.com/v2/url?u=http-3A__www.marist.edu_htbin_wlvindex-3FLINUX-2D390&d=DwIFAg&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=zIUVOGBLCjZgWnU3DmF7lZT7l1bWZgu2QxvNY7dvZdo&s=aBMZ09Pqgp_6XxwrnmjegxtTbjtgS4qyzCWtgPJFh60&e= -- For more information on Linux on System z, visit https://urldefense.proofpoint.com/v2/url?u=http-3A__wiki.linuxvm.org_&d=DwIFAg&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=zIUVOGBLCjZgWnU3DmF7lZT7l1bWZgu2QxvNY7dvZdo&s=PwLIdR5U1lP1jLVBdj1mcn1CUxZcCn7tkQooDnMokUQ&e= -- 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: RHEL 6.9 servers getting Unknown program exception: 0028 SMP
Rick, if you have VM65396 on, but not VM65414, that would be my guess. VM65414 corrected a problem introduced by VM65396 where a guest could erroneously be given a program check 28. ___ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- 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: RHEL 6.9 servers getting Unknown program exception: 0028 SMP
Greg, For information on vulnerabilities, please see the IBM Z Security Portal. Information on the portal can be found if you page down slightly on https://www.ibm.com/it-infrastructure/z/capabilities/system-integrity Regards, Bill ___ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- 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: RHEL 6.9 servers getting Unknown program exception: 0028 SMP
Rick, Terri, Greg, and others, apologies I should have given out the PTF numbers and a little more background on my earlier append. APAR VM65396 PTF UM34851 is the one that introduced the prog check problem. We discovered the problem February 26th, and marked VM65396 in error on March 1st with VM65414 the PE fix to it. The correction to that error is included in APAR VM65414 PTF UM34853, which closed on March 23rd. For other reasons, both of these APARs are Security/Integrity APARs, so see my earlier append on the IBM Z Security portal. However, the functional problem, other than being introduced by VM65396, is not connected to the security/integrity aspects of these APARs. The functional problem is related to a virtual processor coming out of SIE for a fast path operation, as opposed to the typical normal exit from SIE. (For those unfamiliar with SIE, think of it as dispatching or running a virtual processor). Fast path exits are far more prevalent in a CMS environment than in other guest operating systems, but it is possible. In all cases, fast path exits are a subset, mostly a very small subset, of all exits. There is a scenario where taking the fast path exit can erroneously cause z/VM to present the 028 program check. The error further involves the upper part of a register not being cleared. If this register's upper half contains zeroes, it would not trigger the error condition, but over time, it appears the upper half changes and the program checks start appearing. So from that perspective it's a timing/workload dependent problem. This is a simplification, but I hope it's enough for you to appreciate the aspects you need to be aware. The real solution is to apply VM65414's PTF which is available. A mitigation would be to do a CP TRACE PROG 28 NOTERM NOPRINT RUN for the virtual machines through the console or if logged off and on, through a directory command statement. Thank you for understanding the communication challenges in this space. I'm sure you'll let us know if you need more info. :-) Regards, Bill _______ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- 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: RHEL 6.9 servers getting Unknown program exception: 0028 SMP
Rick, As we discussed on phone, but for benefit of others. My last append forgot to point out that the TRACE needs to be done for all the virtual processors in the virtual machine. Otherwise only, only the base VMDBK is protected. This can be done by using the "CPU" command: CPU ALL CMD TRACE PROG 28 NOTERM NOPRINT RUN I apologize for giving both of us a fright and for those following along. Regards, Bill _______ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- 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: Planning for zLINUX guest on zVM 7.1
Peter, It will really depend on the workload and other configuration aspects and what SLA requirements exist. You could start with the degree of over commit. From a memory perspective if you say 2 to 1 or 1.5 to 1, then you get 224G or 168G to your 112G real. It wasn't clear to me if you have 2 real IFL cores or something different. With three logical partitions at logical 2-ways, you'd be at 3:1 there, which is a little higher than I would recommend. The overcommit on virtual to logical processors doesn't have as clear a rule of thumb as memory. So it really depends. Others may be able to offer their experiences. _______ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Technical Linux on IBM Z and LinuxONE Customer Workshop
Technical Linux on IBM Z and LinuxONE Customer Workshop in IBM Poughkeepsie, NY, October 28-29, 2019! Registration extended until October 23! Register here www.ibm.com/de-de/marketing/linuxworkshop-2019/registration.html Agenda highlights IBM z15 Performance Proof Points Introduction and Positioning of Red Hat OpenShift Virtualization Options for Linux on IBM Z and LinuxONE Welcome to the Jungle - Linux on IBM Z Networking Options Z as a Service - Securing Workloads in the Public Cloud Linux on Z Crypto Update How You Can Containerize Your Applications z/VM Platform Update Leveraging the Newest Capabilities in z/VM Interactive Sessions Various Workgroup Sessions Pervasive Encryption Hands-on Session z/VM Problem Determination Demo Session Who should attend IT Architects Application Developers System Administrator System Developers Business partners interested in acquiring Linux on IBM Z skills Start of event: October 28, 2019 at 10:00 a.m. EDT End of event: October 29, 2019 at 4:00 p.m. EDT Where: IBM Poughkeepsie, Client Experience Center, Poughkeepsie, NY, US ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: Technical Linux on IBM Z and LinuxONE Customer Workshop
Unfortunately, no webex for anyone at this time. However, a number of the presentations are available at the following - https://developer.ibm.com/tv/linux-ibm-z/ http://www.vm.ibm.com/library/presentations/ Also, some of these were recently done in Brasilia and the intent is to do more there in the future. Contact Carlos Afonso (carlo...@br.ibm.com) if you are not already on his list for the Linux on LinuxONE workshops he organizes. Regards, Bill ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From: Rinaldo Akio Uehara To: LINUX-390@VM.MARIST.EDU Date: 10/20/2019 02:26 AM Subject:[EXTERNAL] Re: Technical Linux on IBM Z and LinuxONE Customer Workshop Sent by:Linux on 390 Port No live transmission to those outside USA? Enviado do meu iPhone > Em 19 de out de 2019, à(s) 14:04, Bill Bitner escreveu: > > Technical Linux on IBM Z and LinuxONE Customer Workshop in IBM > Poughkeepsie, NY, October 28-29, 2019! > Registration extended until October 23! > Register here > www.ibm.com/de-de/marketing/linuxworkshop-2019/registration.html > > > Agenda highlights > IBM z15 Performance Proof Points > Introduction and Positioning of Red Hat OpenShift > Virtualization Options for Linux on IBM Z and LinuxONE > Welcome to the Jungle - Linux on IBM Z Networking Options > Z as a Service - Securing Workloads in the Public Cloud > Linux on Z Crypto Update > How You Can Containerize Your Applications > z/VM Platform Update > Leveraging the Newest Capabilities in z/VM > Interactive Sessions > Various Workgroup Sessions > Pervasive Encryption Hands-on Session > z/VM Problem Determination Demo Session > > Who should attend > IT Architects > Application Developers > System Administrator > System Developers > Business partners interested in acquiring Linux on IBM Z skills > > Start of event: October 28, 2019 at 10:00 a.m. EDT > End of event: October 29, 2019 at 4:00 p.m. EDT > Where: IBM Poughkeepsie, Client Experience Center, Poughkeepsie, NY, US > ___ > > Bill Bitner - z/VM Client Focus and Care - 607-429-3286 > bitn...@us.ibm.com > "Making systems practical and profitable for customers through > virtualization and its exploitation." - z/VM > > -- > For LINUX-390 subscribe / signoff / archive access instructions, > send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit > https://urldefense.proofpoint.com/v2/url?u=http-3A__www2.marist.edu_htbin_wlvindex-3FLINUX-2D390&d=DwIFaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=yXGrxWuyHIeI-EAhHv_9Pxv1eSu1-_Dn3XkkJZEZazU&s=4UfJxTgbs_umEDvAsrLKDutXr2-JqDFGoqo-Mb06HJc&e= - "Esta mensagem do SERVIÇO FEDERAL DE PROCESSAMENTO DE DADOS (SERPRO), empresa pública federal regida pelo disposto na Lei Federal nº 5.615, é enviada exclusivamente a seu destinatário e pode conter informações confidenciais, protegidas por sigilo profissional. Sua utilização desautorizada é ilegal e sujeita o infrator às penas da lei. Se você a recebeu indevidamente, queira, por gentileza, reenviá-la ao emitente, esclarecendo o equívoco." "This message from SERVIÇO FEDERAL DE PROCESSAMENTO DE DADOS (SERPRO) -- a government company established under Brazilian law (5.615/70) -- is directed exclusively to its addressee and may contain confidential data, protected under professional secrecy rules. Its unauthorized use is illegal and may subject the transgressor to the law's penalties. If you're not the addressee, please send it back, elucidating the failure." -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit https://urldefense.proofpoint.com/v2/url?u=http-3A__www2.marist.edu_htbin_wlvindex-3FLINUX-2D390&d=DwIFaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=yXGrxWuyHIeI-EAhHv_9Pxv1eSu1-_Dn3XkkJZEZazU&s=4UfJxTgbs_umEDvAsrLKDutXr2-JqDFGoqo-Mb06HJc&e= -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: Free Mainframe Stuff 2020: Reply Here with Nominations
>From a z/VM perspective, there is a lot, but a few of the key ones that come to mind https://www.vm.ibm.com/download/packages/ - IBM z/VM Download Packages https://www.vm.ibm.com/library/ - IBM z/VM documentation, presentations, and more http://www.vmworkshop.org/tools.shtml - VM Workshop Tools page https://www.youtube.com/channel/UCwHDyL91yiybsY71dwQpveg - VM Workshop YouTube channel for z/VM, Linux, and z/VSE ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: z15 on-board compression
Thanks Andreas, The problem is there is a bug in WAS when DFLTCC is used - Assessment by Java Development is that Wells Fargo's symptoms line up to be the same issue as experienced by another client several days earlier. In the first client's case, they were provided with an WAS Ifix (APAR PH27505) which prevents the endless looping that occurs when the JVM API does not report that the compressed data stream has reached the end. The client tested the WAS iFix and confirmed that it does prevent the endless looping during decompression. The current recommendation is for z15 zLinux clients running JAVA8 SR6 is to apply WAS APAR PH27505 and run with HW compression disabled (DFLTCC=0). Wells Fargo has 2000 Linux guests they would have to patch to fix and were looking for an easy way to just disable. Regards, Bill _______ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From: Andreas Krebbel To: LINUX-390@VM.MARIST.EDU Date: 07/22/2020 07:49 AM Subject:[EXTERNAL] Re: z15 on-board compression Sent by:Linux on 390 Port Hi, I wrote a post about how to build and test zlib also for distros currently not supported: https://urldefense.proofpoint.com/v2/url?u=http-3A__linux-2Don-2Dz.blogspot.com_2019_10_howto-2Dexploiting-2Dhardware-2Dcompression.html&d=DwICaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=x1vmZzvvtIGmEi8_Y7og3XDpzRz1wa5VLMnZ-0HDnMA&s=uQF9dwXC8luEMIVM-xxLnLikZ8q2yREoz_7unCt3O30&e= You will find a quick one-liner test at the end of the post which can be used to verify whether hardware compression works in your installation. Andreas On 10.06.20 01:36, Michael MacIsaac wrote: > Hello list, > > I heard about the new DFLTCC instruction on the z15, aka on board > compression. I tried a quick experiment to see the difference from a z14. > Disclaimer: I am not a performance expert.> > Here are three commands to create, compress and decompress a 1G file on a > z14: > > # grep Type: /proc/sysinfo > Type: 3906 > > # time dd if=/dev/zero of=1G.file bs=1G count=1 > 1+0 records in > 1+0 records out > 1073741824 bytes (1.1 GB, 1.0 GiB) copied, 21.93 s, 49.0 MB/s > > real0m22.047s > user0m0.001s > sys 0m3.669s > > # time cat 1G.file | gzip -c > 1G.compressed.file > > real0m7.603s > user0m5.362s > sys 0m0.789s > > # time cat 1G.compressed.file | gzip -d > 1G.file > > real0m24.833s > user0m4.103s > sys 0m1.845s > > Here's the same commands on z15: > > # grep Type: /proc/sysinfo > Type: 8561 > > # time dd if=/dev/zero of=1G.file bs=1G count=1 > 1+0 records in > 1+0 records out > 1073741824 bytes (1.1 GB, 1.0 GiB) copied, 1.59126 s, 675 MB/s > > real0m1.621s > user0m0.000s > sys 0m1.216s > > # time cat 1G.file | gzip -c > 1G.compressed.file > > real0m5.722s > user0m4.946s > sys 0m0.510s > # time cat 1G.compressed.file | gzip -d > 1G.file > > real0m6.150s > user0m3.922s > sys 0m1.290s > > Wow more than 10x faster on dd - was not expecting that as I didn't think > it uses compression. But the compress with gzip -c, was only 25% faster on > the z15 while the decompress was about 4x. > > Are these results expected? > > Thanks. > > > -- > -Mike MacIsaac > > -- > For LINUX-390 subscribe / signoff / archive access instructions, > send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit > https://urldefense.proofpoint.com/v2/url?u=http-3A__www2.marist.edu_htbin_wlvindex-3FLINUX-2D390&d=DwICaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=x1vmZzvvtIGmEi8_Y7og3XDpzRz1wa5VLMnZ-0HDnMA&s=FQsLX2gZ60YhLotQ1szspOspNp8arGQloNna6pa4DhI&e= > -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit https://urldefense.proofpoint.com/v2/url?u=http-3A__www2.marist.edu_htbin_wlvindex-3FLINUX-2D390&d=DwICaQ&c=jf_iaSHvJObTbx-siA1ZOg&r=5MoiGnRMsqjUxHq7C7RX4kthfAKqf40IRGxojgucrwA&m=x1vmZzvvtIGmEi8_Y7og3XDpzRz1wa5VLMnZ-0HDnMA&s=FQsLX2gZ60YhLotQ1szspOspNp8arGQloNna6pa4DhI&e= -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: z15 on-board compression
Oops ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: vmrelocate and quiescence time
I take a few days off and a fun topic comes up that I mostly miss. :-) We did a Live Virtual Class on this back when SSI first came out. A few things have changed, but you may find value in it: http://www.vm.ibm.com/education/lvc/zvmlvc.html The charts for the presentation are http://www.vm.ibm.com/education/lvc/LVC0227.pdf It goes through the various factors that influence overall relocation time and quiesce time. What we find with databases is often they are large virtual machines in terms of memory. This can can have a non-trivial effect on quiesce time as the process involves traversing the DAT structures to validate changed pages. This traversal is done twice for various reasons. So chart 23 of the presentation deck illustrates that effect. This has to be done regardless of whether pages change or there is activity because this is the mechanism used to determine that. Since that LVC there were updates to Performance Toolkit that can show some of the statistics on LGR (or in other z/VM Performance products). ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: vmrelocate and quiescence time
z/VM doesn't have a short cut to determine that no pages have changed. So for a guest over 100GB, it has to examine over 26 million things multiple times. Validating the I/O is drained is another aspect, but my guess is the traversing of the DAT structures is the biggest factor. Regards, Bill ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From: Grzegorz Powiedziuk To: LINUX-390@VM.MARIST.EDU Date: 08/17/2020 11:51 PM Subject:[EXTERNAL] Re: vmrelocate and quiescence time Sent by:Linux on 390 Port On Mon, Aug 17, 2020 at 12:26 PM Bill Bitner wrote: > I take a few days off and a fun topic comes up that I mostly miss. :-) > > We did a Live Virtual Class on this back when SSI first came out. A few > things have changed, but you may find value in it: > http://www.vm.ibm.com/education/lvc/zvmlvc.html > The charts for the presentation are > http://www.vm.ibm.com/education/lvc/LVC0227.pdf > > This is great! Thank you. Although this brings a few more questions, it was definitely very helpful. - chart on page 24 idle case (0GB change) had a quiesce time 8s - I wonder why. That VM in this test was probably >100GB so more or less close to my case. But I thought that if there are close to 0 changes, then the last pass should be just a "formality"? I was relocating my VM with DB2 being stopped! There were just few idle processes left and yet it took 20-30s of freeze time. Of course I am doing sync and not immediate. Seems like I should be below <10s even with just 2CTCs (although I am hoping to get 2 more) by looking at these charts. - document says that I/O device count matters as well. We have 4 FCP channels with quite a big number of luns (>20), would that make a big difference? I will look into anything related in perfsvm thanks again. Gregory -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: vmrelocate and quiescence time
There have been different things discussed, though I don't recall if guarded storage facility was one of them (actually introduced in z14) as that came out after we did the initial LGR. ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From: Neale Ferguson To: LINUX-390@VM.MARIST.EDU Date: 08/18/2020 10:04 AM Subject:[EXTERNAL] Re: vmrelocate and quiescence time Sent by:Linux on 390 Port Could you take advantage of the guard page hardware facility of the z15 that the pause-less Java garbage collector uses? Original message z/VM doesn't have a short cut to determine that no pages have changed. So for a guest over 100GB, it has to examine over 26 million things multiple times. Validating the I/O is drained is another aspect, but my guess is the traversing of the DAT structures is the biggest factor. -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Crowdstrike Interest
Hello, asking for a friend.:-) Does anyone have interest in the the end point security products from Crowdstrike being supported on IBM Z and LinuxONE? The IBM team that works with 3rd parties is looking to gather information on interest. I know they have heard from some of you, but I wanted to see if others had interest. Thanks everyone. You can reply here or email bitn...@us.ibm.com. Regards, Bill ___ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: IFL usage of Linux servers
Hello David, An important aspect of this is whether you are running SMT-2 or not. If you are, then I recommend you read https://www.vm.ibm.com/perf/tips/smtutil.html for some background as when SMT-2 is on we have to take extra care because in that scenario the word "IFL" is ambiguous between whether we're referencing the Core or the two threads in aggregate, as with SMT-2 many of the measures, but not all, are of a logical processor (aka thread) perspective. If you share whether you're SMT-2 or SMIT-1 or SMT-0 (SMT disabled). I can give you some additional information on interpreting the data. Regards, Bill _______ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From: "Mittelstädt, David" To: LINUX-390@VM.MARIST.EDU Date: 10/28/2020 01:30 PM Subject:[EXTERNAL] IFL usage of Linux servers Sent by:Linux on 390 Port Hi everyone, I need a little help understanding the values from the z/VM Performance Toolkit and command line tool hyptop. I want to measure the IFL utilization of a z/VM guest running linux from a z/VM or hardware perspective, e.g. linux server A consumes 2 IFL or linux server B consumes 0.5 IFL. We need this information for accounting and forecast purposes. So I used the tool hyptop to get the information of the used cpu, but I'm not sure how to interpret these values correctly. What means the column cpu(%)? Is this the usage of 1 IFL in percent? And is this same value like the one in the z/VM Performance Toolkit %CPU under 21. User resource usage? Is it possible to measure the IFL usage with these values? Or do I need other values? Thanks in advance. Best regards, David Mittelstädt -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: AW: [EXTERN]-Re: IFL usage of Linux servers
Ok. So then the data you'll see in Hytop when on z/VM and in some of the Performance Toolkit reports (e.g. FCX112 USER) will show utilization of a logical processor (aka thread), so there isn't a direct correlation to the total Core Utilization from that perspective. Hytop will give you "CPU" time, again think of this as thread time of one of the 2 threads per Core. Hytop breaks down what it calls Total CPU time and then Total Management time. The "Management" time will relate to time spent in z/VM hypervisor or Control Program. So on the FCX112 report, z/VM shows Total and Emul (short for Emulation or Virtual). Emul would be time spent in Linux and its applications and the delta of Total and Emul from Performance Toolkit on FCX112 will map closer to that "Management" time from Hytop. There is an APAR coming for Performance Toolkit that on z/VM 7.2 (VM66215) that will surface some other counters related to processor usage in SMT-2 environment. I hope that helps. That URL will give some more details. _______ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From: "Mittelstädt, David" To: LINUX-390@VM.MARIST.EDU Date: 10/28/2020 02:29 PM Subject:[EXTERNAL] AW: [EXTERN]-Re: IFL usage of Linux servers Sent by:Linux on 390 Port Hi Bill, Thanks for the fast replay. I will read the infos in the given URL tomorrow. We are using SMT-2. I ran a q multithreading in z/VM and get back Activated Threads 2 for IFL Core. I'm really interested in the additional information. Thanks. Regards, David -Ursprüngliche Nachricht- Von: Linux on 390 Port [mailto:LINUX-390@VM.MARIST.EDU] Im Auftrag von Bill Bitner Gesendet: Mittwoch, 28. Oktober 2020 19:05 An: LINUX-390@VM.MARIST.EDU Betreff: [EXTERN]-Re: IFL usage of Linux servers Hello David, An important aspect of this is whether you are running SMT-2 or not. If you are, then I recommend you read https://www.vm.ibm.com/perf/tips/smtutil.html for some background as when SMT-2 is on we have to take extra care because in that scenario the word "IFL" is ambiguous between whether we're referencing the Core or the two threads in aggregate, as with SMT-2 many of the measures, but not all, are of a logical processor (aka thread) perspective. If you share whether you're SMT-2 or SMIT-1 or SMT-0 (SMT disabled). I can give you some additional information on interpreting the data. Regards, Bill _______ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM From:"Mittelstädt, David" To: LINUX-390@VM.MARIST.EDU Date:10/28/2020 01:30 PM Subject: [EXTERNAL] IFL usage of Linux servers Sent by: Linux on 390 Port Hi everyone, I need a little help understanding the values from the z/VM Performance Toolkit and command line tool hyptop. I want to measure the IFL utilization of a z/VM guest running linux from a z/VM or hardware perspective, e.g. linux server A consumes 2 IFL or linux server B consumes 0.5 IFL. We need this information for accounting and forecast purposes. So I used the tool hyptop to get the information of the used cpu, but I'm not sure how to interpret these values correctly. What means the column cpu(%)? Is this the usage of 1 IFL in percent? And is this same value like the one in the z/VM Performance Toolkit %CPU under 21. User resource usage? Is it possible to measure the IFL usage with these values? Or do I need other values? Thanks in advance. Best regards, David Mittelstädt -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390 -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: Question About LGR and Hipersocket using RHEL 7.7
Larry, If this isn't resolved, I'd suggest you open a z/VM support case and the team and take a closer look. Thanks, Bill _______ Bill Bitner - z/VM Client Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Redbooks
I do my best to stay out of the rumor business. Chris Konarski just tweeted that is is a misunderstanding. https://twitter.com/ChrisKonarski/status/1454110707096637442 Chris is the IBM VP WW Technical Sales and Lab Services. Have a nice weekend everyone. Regards, Bill "Not speaking for IBM as far as they know" _______ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
Re: IBM vs other virtualizations
Hello Carey, Of course I remember you, I just haven't checked the list in a few days. But as you saw our many friends did! I am not aware of a good resource for your question. The challenge is hardly no one knows the virtualization methods on x86 and the virtualization methods on IBM Z (or Power) in enough detail to contrast them. Typically one knows one or two of them in detail, but seldom all of them. The other challenge is these aren't static systems, everyone is making enhancements and some of the comparisons out there are for a point in time, or in some cases IBM in 1990 compared to x86 in 2020 (as if the mainframe hasn't changed). Sorry I don't have better pointers. Regards, Bill _______ Bill Bitner - z/VM Customer Focus and Care - 607-429-3286 bitn...@us.ibm.com "Making systems practical and profitable for customers through virtualization and its exploitation." - z/VM -- For LINUX-390 subscribe / signoff / archive access instructions, send email to lists...@vm.marist.edu with the message: INFO LINUX-390 or visit http://www2.marist.edu/htbin/wlvindex?LINUX-390
