Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances
let's say you have a RRT of 180 ms what you then need is your theoretical link speed - let's say 10 Gbit/s ... easily let's take 1 GB/sthis means, you socket must be capable to take your bandwidth (data stream) during the "first" 180ms because it will take at least this time to get back the first ACKs .. .so 1 GB / s x 0,180 s = 1024 MB/s x 0,180 s ==>> 185 MB this means, you have to allow the operating system to accept socketsizes in that range... set something like this - but increase these values to 185 MBsysctl -w net.ipv4.tcp_rmem="12194304 12194304 12194304" sysctl -w net.ipv4.tcp_wmem="12194304 12194304 12194304"sysctl -w net.ipv4.tcp_mem="12194304 12194304 12194304"sysctl -w net.core.rmem_max=12194304sysctl -w net.core.wmem_max=12194304sysctl -w net.core.rmem_default=12194304sysctl -w net.core.wmem_default=12194304sysctl -w net.core.optmem_max=12194304in addition set this :sysctl -w net.core.netdev_max_backlog=5sysctl -w net.ipv4.tcp_no_metrics_save=1sysctl -w net.ipv4.tcp_timestamps=0sysctl -w net.ipv4.tcp_sack=1sysctl -w net.core.netdev_max_backlog=5sysctl -w net.ipv4.tcp_max_syn_backlog=3you need to "recycle" the sockets.. means .. mmshutdown/stsartuposhould fix you issueMit freundlichen Grüßen / Kind regards Olaf Weiser EMEA Storage Competence Center Mainz, German / IBM Systems, Storage Platform,---IBM DeutschlandIBM Allee 171139 EhningenPhone: +49-170-579-44-66E-Mail: olaf.wei...@de.ibm.com---IBM Deutschland GmbH / Vorsitzender des Aufsichtsrats: Martin JetterGeschäftsführung: Martina Koederitz (Vorsitzende), Susanne Peter, Norbert Janzen, Dr. Christian Keller, Ivo Koerner, Markus KoernerSitz der Gesellschaft: Ehningen / Registergericht: Amtsgericht Stuttgart, HRB 14562 / WEEE-Reg.-Nr. DE 99369940 From: Jan-Frode MyklebustTo: "gpfsug-discuss@spectrumscale.org" Date: 11/09/2016 07:05 PMSubject: Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distancesSent by: gpfsug-discuss-boun...@spectrumscale.orgMostly curious, don't have experience in such environments, but ... Is this AFM over NFS or NSD protocol? Might be interesting to try the other option -- and also check how nsdperf performs over such distance/latency.-jfons. 9. nov. 2016 kl. 18.39 skrev Jake Carroll :Hi. I’ve got an GPFS to GPFS AFM cache/home (IW) relationship set up over a really long distance. About 180ms of latency between the two clusters and around 13,000km of optical path. Fortunately for me, I’ve actually got near theoretical maximum IO over the NIC’s between the clusters and I’m iPerf’ing at around 8.90 to 9.2Gbit/sec over a 10GbE circuit. All MTU9000 all the way through. Anyway – I’m finding my AFM traffic to be dragging its feet and I don’t really understand why that might be. I’ve verified the links and transports ability as I said above with iPerf, and CERN’s FDT to near 10Gbit/sec. I also verified the clusters on both sides in terms of disk IO and they both seem easily capable in IOZone and IOR tests of multiple GB/sec of throughput. So – my questions: 1. Are there very specific tunings AFM needs for high latency/long distance IO? 2. Are there very specific NIC/TCP-stack tunings (beyond the type of thing we already have in place) that benefits AFM over really long distances and high latency?3. We are seeing on the “cache” side really lazy/sticky “ls –als” in the home mount. It sometimes takes 20 to 30 seconds before the command line will report back with a long listing of files. Any ideas why it’d take that long to get a response from “home”. We’ve got our TCP stack setup fairly aggressively, on all hosts that participate in these two clusters. ethtool -C enp2s0f0 adaptive-rx offifconfig enp2s0f0 txqueuelen 1sysctl -w net.core.rmem_max=536870912sysctl -w net.core.wmem_max=536870912sysctl -w net.ipv4.tcp_rmem="4096 87380 268435456"sysctl -w net.ipv4.tcp_wmem="4096 65536 268435456"sysctl -w net.core.netdev_max_backlog=25sysctl -w net.ipv4.tcp_congestion_control=htcpsysctl -w net.ipv4.tcp_mtu_probing=1 I modified a couple of small things on the AFM “cache” side to see if it’d make a difference such as: mmchconfig afmNumWriteThreads=4mmchconfig afmNumReadThreads=4 But no difference so far. Thoughts would be appreciated. I’ve done this before over much shorter distances (30Km) and I’ve flattened a 10GbE wire without really tuning…anything. Are my large in-flight-packets numbers/long-time-to-acknowledgement semantics going to hurt here? I really thought AFM might be well designed for exactly this kind of work at long distance *and* high throughput – so I must
Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances (Jan-Frode Myklebust)
So you are using the NSD protocol for data transfers over multi-cluster? If so the TCP and thread tuning should help as well. Scott Fadden Spectrum Scale - Technical Marketing Phone: (503) 880-5833 sfad...@us.ibm.com http://www.ibm.com/systems/storage/spectrum/scale From: Jake Carroll <jake.carr...@uq.edu.au> To: "gpfsug-discuss@spectrumscale.org" <gpfsug-discuss@spectrumscale.org> Date: 11/09/2016 10:09 AM Subject: Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances (Jan-Frode Myklebust) Sent by:gpfsug-discuss-boun...@spectrumscale.org Hi jf… >>Mostly curious, don't have experience in such environments, but ... Is this AFM over NFS or NSD protocol? Might be interesting to try the other option -- and also check how nsdperf performs over such distance/latency. As it turns out, it seems, very few people do. I will test nsdperf over it and see how it performs. And yes, it is AFM → AFM. No NFS involved here! -jc -- Message: 2 Date: Wed, 9 Nov 2016 17:39:05 + From: Jake Carroll <jake.carr...@uq.edu.au> To: "gpfsug-discuss@spectrumscale.org" <gpfsug-discuss@spectrumscale.org> Subject: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances Message-ID: <83652c3d-0802-4cc2-b636-9faa31ef5...@uq.edu.au> Content-Type: text/plain; charset="utf-8" Hi. I?ve got an GPFS to GPFS AFM cache/home (IW) relationship set up over a really long distance. About 180ms of latency between the two clusters and around 13,000km of optical path. Fortunately for me, I?ve actually got near theoretical maximum IO over the NIC?s between the clusters and I?m iPerf?ing at around 8.90 to 9.2Gbit/sec over a 10GbE circuit. All MTU9000 all the way through. Anyway ? I?m finding my AFM traffic to be dragging its feet and I don?t really understand why that might be. I?ve verified the links and transports ability as I said above with iPerf, and CERN?s FDT to near 10Gbit/sec. I also verified the clusters on both sides in terms of disk IO and they both seem easily capable in IOZone and IOR tests of multiple GB/sec of throughput. So ? my questions: 1. Are there very specific tunings AFM needs for high latency/long distance IO? 2. Are there very specific NIC/TCP-stack tunings (beyond the type of thing we already have in place) that benefits AFM over really long distances and high latency? 3. We are seeing on the ?cache? side really lazy/sticky ?ls ?als? in the home mount. It sometimes takes 20 to 30 seconds before the command line will report back with a long listing of files. Any ideas why it?d take that long to get a response from ?home?. We?ve got our TCP stack setup fairly aggressively, on all hosts that participate in these two clusters. ethtool -C enp2s0f0 adaptive-rx off ifconfig enp2s0f0 txqueuelen 1 sysctl -w net.core.rmem_max=536870912 sysctl -w net.core.wmem_max=536870912 sysctl -w net.ipv4.tcp_rmem="4096 87380 268435456" sysctl -w net.ipv4.tcp_wmem="4096 65536 268435456" sysctl -w net.core.netdev_max_backlog=25 sysctl -w net.ipv4.tcp_congestion_control=htcp sysctl -w net.ipv4.tcp_mtu_probing=1 I modified a couple of small things on the AFM ?cache? side to see if it?d make a difference such as: mmchconfig afmNumWriteThreads=4 mmchconfig afmNumReadThreads=4 But no difference so far. Thoughts would be appreciated. I?ve done this before over much shorter distances (30Km) and I?ve flattened a 10GbE wire without really tuning?anything. Are my large in-flight-packets numbers/long-time-to-acknowledgement semantics going to hurt here? I really thought AFM might be well designed for exactly this kind of work at long distance *and* high throughput ? so I must be missing something! -jc -- next part -- An HTML attachment was scrubbed... URL: < http://gpfsug.org/pipermail/gpfsug-discuss/attachments/20161109/d4f4d9a7/attachment-0001.html > -- Message: 3 Date: Wed, 09 Nov 2016 18:05:21 + From: Jan-Frode Myklebust <janfr...@tanso.net> To: "gpfsug-discuss@spectrumscale.org" <gpfsug-discuss@spectrumscale.org> Subject: Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances Message-ID:
Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances (Jan-Frode Myklebust)
Hi jf… >>Mostly curious, don't have experience in such environments, but ... Is >> this AFM over NFS or NSD protocol? Might be interesting to try the other option -- and also check how nsdperf performs over such distance/latency. As it turns out, it seems, very few people do. I will test nsdperf over it and see how it performs. And yes, it is AFM → AFM. No NFS involved here! -jc -- Message: 2 Date: Wed, 9 Nov 2016 17:39:05 + From: Jake Carroll <jake.carr...@uq.edu.au> To: "gpfsug-discuss@spectrumscale.org" <gpfsug-discuss@spectrumscale.org> Subject: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances Message-ID: <83652c3d-0802-4cc2-b636-9faa31ef5...@uq.edu.au> Content-Type: text/plain; charset="utf-8" Hi. I?ve got an GPFS to GPFS AFM cache/home (IW) relationship set up over a really long distance. About 180ms of latency between the two clusters and around 13,000km of optical path. Fortunately for me, I?ve actually got near theoretical maximum IO over the NIC?s between the clusters and I?m iPerf?ing at around 8.90 to 9.2Gbit/sec over a 10GbE circuit. All MTU9000 all the way through. Anyway ? I?m finding my AFM traffic to be dragging its feet and I don?t really understand why that might be. I?ve verified the links and transports ability as I said above with iPerf, and CERN?s FDT to near 10Gbit/sec. I also verified the clusters on both sides in terms of disk IO and they both seem easily capable in IOZone and IOR tests of multiple GB/sec of throughput. So ? my questions: 1. Are there very specific tunings AFM needs for high latency/long distance IO? 2. Are there very specific NIC/TCP-stack tunings (beyond the type of thing we already have in place) that benefits AFM over really long distances and high latency? 3. We are seeing on the ?cache? side really lazy/sticky ?ls ?als? in the home mount. It sometimes takes 20 to 30 seconds before the command line will report back with a long listing of files. Any ideas why it?d take that long to get a response from ?home?. We?ve got our TCP stack setup fairly aggressively, on all hosts that participate in these two clusters. ethtool -C enp2s0f0 adaptive-rx off ifconfig enp2s0f0 txqueuelen 1 sysctl -w net.core.rmem_max=536870912 sysctl -w net.core.wmem_max=536870912 sysctl -w net.ipv4.tcp_rmem="4096 87380 268435456" sysctl -w net.ipv4.tcp_wmem="4096 65536 268435456" sysctl -w net.core.netdev_max_backlog=25 sysctl -w net.ipv4.tcp_congestion_control=htcp sysctl -w net.ipv4.tcp_mtu_probing=1 I modified a couple of small things on the AFM ?cache? side to see if it?d make a difference such as: mmchconfig afmNumWriteThreads=4 mmchconfig afmNumReadThreads=4 But no difference so far. Thoughts would be appreciated. I?ve done this before over much shorter distances (30Km) and I?ve flattened a 10GbE wire without really tuning?anything. Are my large in-flight-packets numbers/long-time-to-acknowledgement semantics going to hurt here? I really thought AFM might be well designed for exactly this kind of work at long distance *and* high throughput ? so I must be missing something! -jc -- next part -- An HTML attachment was scrubbed... URL: <http://gpfsug.org/pipermail/gpfsug-discuss/attachments/20161109/d4f4d9a7/attachment-0001.html> -- Message: 3 Date: Wed, 09 Nov 2016 18:05:21 + From: Jan-Frode Myklebust <janfr...@tanso.net> To: "gpfsug-discuss@spectrumscale.org" <gpfsug-discuss@spectrumscale.org> Subject: Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances Message-ID:
Re: [gpfsug-discuss] Tuning AFM for high throughput/high IO over _really_ long distances
Jake, If AFM is using NFS it is all about NFS tuning. The copy from one side to the other is basically just a client writing to an NFS mount. Thee are a few things you can look at: 1. NFS Transfer size (Make is 1MiB, I think that is the max) 2. TCP Tuning for large window size. This is discussed on Tuning active file management home communications in the docs. On this page you will find some discussion on increasing gateway threads, and other things similar that may help as well. We can discuss further as I understand we will be meeting at SC16. Scott Fadden Spectrum Scale - Technical Marketing Phone: (503) 880-5833 sfad...@us.ibm.com http://www.ibm.com/systems/storage/spectrum/scale From: Jake CarrollTo: "gpfsug-discuss@spectrumscale.org" Date: 11/09/2016 09:39 AM Subject:[gpfsug-discuss] Tuning AFM for high throughput/high IO over_really_ long distances Sent by:gpfsug-discuss-boun...@spectrumscale.org Hi. I’ve got an GPFS to GPFS AFM cache/home (IW) relationship set up over a really long distance. About 180ms of latency between the two clusters and around 13,000km of optical path. Fortunately for me, I’ve actually got near theoretical maximum IO over the NIC’s between the clusters and I’m iPerf’ing at around 8.90 to 9.2Gbit/sec over a 10GbE circuit. All MTU9000 all the way through. Anyway – I’m finding my AFM traffic to be dragging its feet and I don’t really understand why that might be. I’ve verified the links and transports ability as I said above with iPerf, and CERN’s FDT to near 10Gbit/sec. I also verified the clusters on both sides in terms of disk IO and they both seem easily capable in IOZone and IOR tests of multiple GB/sec of throughput. So – my questions: 1. Are there very specific tunings AFM needs for high latency/long distance IO? 2. Are there very specific NIC/TCP-stack tunings (beyond the type of thing we already have in place) that benefits AFM over really long distances and high latency? 3. We are seeing on the “cache” side really lazy/sticky “ls –als” in the home mount. It sometimes takes 20 to 30 seconds before the command line will report back with a long listing of files. Any ideas why it’d take that long to get a response from “home”. We’ve got our TCP stack setup fairly aggressively, on all hosts that participate in these two clusters. ethtool -C enp2s0f0 adaptive-rx off ifconfig enp2s0f0 txqueuelen 1 sysctl -w net.core.rmem_max=536870912 sysctl -w net.core.wmem_max=536870912 sysctl -w net.ipv4.tcp_rmem="4096 87380 268435456" sysctl -w net.ipv4.tcp_wmem="4096 65536 268435456" sysctl -w net.core.netdev_max_backlog=25 sysctl -w net.ipv4.tcp_congestion_control=htcp sysctl -w net.ipv4.tcp_mtu_probing=1 I modified a couple of small things on the AFM “cache” side to see if it’d make a difference such as: mmchconfig afmNumWriteThreads=4 mmchconfig afmNumReadThreads=4 But no difference so far. Thoughts would be appreciated. I’ve done this before over much shorter distances (30Km) and I’ve flattened a 10GbE wire without really tuning…anything. Are my large in-flight-packets numbers/long-time-to-acknowledgement semantics going to hurt here? I really thought AFM might be well designed for exactly this kind of work at long distance *and* high throughput – so I must be missing something! -jc ___ gpfsug-discuss mailing list gpfsug-discuss at spectrumscale.org http://gpfsug.org/mailman/listinfo/gpfsug-discuss ___ gpfsug-discuss mailing list gpfsug-discuss at spectrumscale.org http://gpfsug.org/mailman/listinfo/gpfsug-discuss