Hi Kamil, in my mail just a few minutes ago I'd overlooked that the buffer size in your trace was indeed 128M (I suppose the trace file is adapting that size if not set in particular). That is very strange, even under high load, the trace should then capture some longer time than 10 secs, and , most of all, it should contain much more activities than just these few you had. That is very mysterious. I am out of ideas for the moment, and a bit short of time to dig here.
To check your tracing, you could run a trace like before but when everything is normal and check that out - you should see many records, the trcsum.awk should list just a small portion of unfinished ops at the end, ... If that is fine, then the tracing itself is affected by your crritical condition (never experienced that before - rather GPFS grinds to a halt than the trace is abandoned), and that might well be worth a support ticket. Mit freundlichen Grüßen / Kind regards Dr. Uwe Falke IT Specialist Hybrid Cloud Infrastructure / Technology Consulting & Implementation Services +49 175 575 2877 Mobile Rathausstr. 7, 09111 Chemnitz, Germany uwefa...@de.ibm.com IBM Services IBM Data Privacy Statement IBM Deutschland Business & Technology Services GmbH Geschäftsführung: Sven Schooss, Stefan Hierl Sitz der Gesellschaft: Ehningen Registergericht: Amtsgericht Stuttgart, HRB 17122 From: Uwe Falke/Germany/IBM To: gpfsug main discussion list <gpfsug-discuss@spectrumscale.org> Date: 13/11/2020 10:21 Subject: Re: [EXTERNAL] Re: [gpfsug-discuss] Poor client performance with high cpu usage of mmfsd process Hi, Kamil, looks your tracefile setting has been too low: all streams included Thu Nov 12 20:58:19.950515266 2020 (TOD 1605232699.950515, cycles 20701552715873212) <---- useful part of trace extends from here trace quiesced Thu Nov 12 20:58:20.133134000 2020 (TOD 1605232700.000133, cycles 20701553190681534) <---- to here means you effectively captured a period of about 5ms only ... you can't see much from that. I'd assumed the default trace file size would be sufficient here but it doesn't seem to. try running with something like mmtracectl --start --trace-file-size=512M --trace=io --tracedev-write-mode=overwrite -N <your_critical_node>. However, if you say "no major waiter" - how many waiters did you see at any time? what kind of waiters were the oldest, how long they'd waited? it could indeed well be that some job is just creating a killer workload. The very short cyle time of the trace points, OTOH, to high activity, OTOH the trace file setting appears quite low (trace=io doesnt' collect many trace infos, just basic IO stuff). If I might ask: what version of GPFS are you running? Mit freundlichen Grüßen / Kind regards Dr. Uwe Falke IT Specialist Hybrid Cloud Infrastructure / Technology Consulting & Implementation Services +49 175 575 2877 Mobile Rathausstr. 7, 09111 Chemnitz, Germany uwefa...@de.ibm.com IBM Services IBM Data Privacy Statement IBM Deutschland Business & Technology Services GmbH Geschäftsführung: Sven Schooss, Stefan Hierl Sitz der Gesellschaft: Ehningen Registergericht: Amtsgericht Stuttgart, HRB 17122 From: "Czauz, Kamil" <kamil.cz...@squarepoint-capital.com> To: gpfsug main discussion list <gpfsug-discuss@spectrumscale.org> Date: 13/11/2020 03:33 Subject: [EXTERNAL] Re: [gpfsug-discuss] Poor client performance with high cpu usage of mmfsd process Sent by: gpfsug-discuss-boun...@spectrumscale.org Hi Uwe - I hit the issue again today, no major waiters, and nothing useful from the iohist report. Nothing interesting in the logs either. I was able to get a trace today while the issue was happening. I took 2 traces a few min apart. The beginning of the traces look something like this: Overwrite trace parameters: buffer size: 134217728 64 kernel trace streams, indices 0-63 (selected by low bits of processor ID) 128 daemon trace streams, indices 64-191 (selected by low bits of thread ID) Interval for calibrating clock rate was 100.019054 seconds and 260049296314 cycles Measured cycle count update rate to be 2599997559 per second <---- using this value OS reported cycle count update rate as 2599999000 per second Trace milestones: kernel trace enabled Thu Nov 12 20:56:40.114080000 2020 (TOD 1605232600.114080, cycles 20701293141385220) daemon trace enabled Thu Nov 12 20:56:40.247430000 2020 (TOD 1605232600.247430, cycles 20701293488095152) all streams included Thu Nov 12 20:58:19.950515266 2020 (TOD 1605232699.950515, cycles 20701552715873212) <---- useful part of trace extends from here trace quiesced Thu Nov 12 20:58:20.133134000 2020 (TOD 1605232700.000133, cycles 20701553190681534) <---- to here Approximate number of times the trace buffer was filled: 553.529 Here is the output of trsum.awk details=0 I'm not quite sure what to make of it, can you help me decipher it? The 'lookup' operations are taking a hell of a long time, what does that mean? Capture 1 Unfinished operations: 21234 ***************** lookup ************** 0.165851604 290020 ***************** lookup ************** 0.151032241 302757 ***************** lookup ************** 0.168723402 301677 ***************** lookup ************** 0.070016530 230983 ***************** lookup ************** 0.127699082 21233 ***************** lookup ************** 0.060357257 309046 ***************** lookup ************** 0.157124551 301643 ***************** lookup ************** 0.165543982 304042 ***************** lookup ************** 0.172513838 167794 ***************** lookup ************** 0.056056815 189680 ***************** lookup ************** 0.062022237 362216 ***************** lookup ************** 0.072063619 406314 ***************** lookup ************** 0.114121838 167776 ***************** lookup ************** 0.114899642 303016 ***************** lookup ************** 0.144491120 290021 ***************** lookup ************** 0.142311603 167762 ***************** lookup ************** 0.144240366 248530 ***************** lookup ************** 0.168728131 0 0.000000000 ********* Unfinished IO: buffer/disk 30018014000 14:48493092752^\00000000FFFFFFFF 0 0.000000000 ********* Unfinished IO: buffer/disk 30018014000 14:48493092752^\FFFFFFFE 0 0.000000000 ********* Unfinished IO: buffer/disk 2000006B000 2:6058336^\00000000FFFFFFFF 0 0.000000000 ********* Unfinished IO: buffer/disk 30018014000 14:48493092752^\FFFFFFFF 0 0.000000000 ********* Unfinished IO: buffer/disk 2000006B000 2:6058336^\FFFFFFFE 0 0.000000000 ********* Unfinished IO: buffer/disk 2000006B000 2:6058336^\FFFFFFFF 0 0.000000000 ********* Unfinished IO: buffer/disk 3000002E000 2:1917676744^\FFFFFFFE 0 0.000000000 ********* Unfinished IO: buffer/disk 3000002E000 2:1917676744^\00000000FFFFFFFF 0 0.000000000 ********* Unfinished IO: buffer/disk 3000002E000 2:1917676744^\FFFFFFFF Elapsed trace time: 0.182617894 seconds Elapsed trace time from first VFS call to last: 0.182617893 Time idle between VFS calls: 0.000006317 seconds Operations stats: total time(s) count avg-usecs wait-time(s) avg-usecs rdwr 0.012021696 35 343.477 read_inode2 0.000100787 43 2.344 follow_link 0.000050609 8 6.326 pagein 0.000097806 10 9.781 revalidate 0.000010884 156 0.070 open 0.001001824 18 55.657 lookup 1.152449696 36 32012.492 delete_inode 0.000036816 38 0.969 permission 0.000080574 14 5.755 release 0.000470096 18 26.116 mmap 0.000340095 9 37.788 llseek 0.000001903 9 0.211 User thread stats: GPFS-time(sec) Appl-time GPFS-% Appl-% Ops 221919 0.000000244 0.050064080 0.00% 100.00% 4 167794 0.000011891 0.000069707 14.57% 85.43% 4 309046 0.147664569 0.000074663 99.95% 0.05% 9 349767 0.000000070 0.000000000 100.00% 0.00% 1 301677 0.017638372 0.048741086 26.57% 73.43% 12 84407 0.000010448 0.000016977 38.10% 61.90% 3 406314 0.000002279 0.000122367 1.83% 98.17% 7 25464 0.043270937 0.000006200 99.99% 0.01% 2 362216 0.000005617 0.000017498 24.30% 75.70% 2 379982 0.000000626 0.000000000 100.00% 0.00% 1 230983 0.123947465 0.000056796 99.95% 0.05% 6 21233 0.047877661 0.004887113 90.74% 9.26% 17 302757 0.154486003 0.010695642 93.52% 6.48% 24 248530 0.000006763 0.000035442 16.02% 83.98% 3 303016 0.014678039 0.000013098 99.91% 0.09% 2 301643 0.088025575 0.054036566 61.96% 38.04% 33 3339 0.000034997 0.178199426 0.02% 99.98% 35 21234 0.164240073 0.000262711 99.84% 0.16% 39 167762 0.000011886 0.000041865 22.11% 77.89% 3 336006 0.000001246 0.100519562 0.00% 100.00% 16 304042 0.121322325 0.019218406 86.33% 13.67% 33 301644 0.054325242 0.087715613 38.25% 61.75% 37 301680 0.000015005 0.020838281 0.07% 99.93% 9 290020 0.147713357 0.000121422 99.92% 0.08% 19 290021 0.000476072 0.000085833 84.72% 15.28% 10 44777 0.040819757 0.000010957 99.97% 0.03% 3 189680 0.000000044 0.000002376 1.82% 98.18% 1 241759 0.000000698 0.000000000 100.00% 0.00% 1 184839 0.000001621 0.150341986 0.00% 100.00% 28 362220 0.000010818 0.000020949 34.05% 65.95% 2 104687 0.000000495 0.000000000 100.00% 0.00% 1 # total App-read/write = 45 Average duration = 0.000269322 sec # time(sec) count % %ile read write avgBytesR avgBytesW 0.000500 34 0.755556 0.755556 34 0 32889 0 0.001000 10 0.222222 0.977778 10 0 108136 0 0.004000 1 0.022222 1.000000 1 0 8 0 # max concurrant App-read/write = 2 # conc count % %ile 1 38 0.844444 0.844444 2 7 0.155556 1.000000 Capture 2 Unfinished operations: 335096 ***************** lookup ************** 0.289127895 334691 ***************** lookup ************** 0.225380797 362246 ***************** lookup ************** 0.052106493 334694 ***************** lookup ************** 0.048567769 362220 ***************** lookup ************** 0.054825580 333972 ***************** lookup ************** 0.275355791 406314 ***************** lookup ************** 0.283219905 334686 ***************** lookup ************** 0.285973208 289606 ***************** lookup ************** 0.064608288 21233 ***************** lookup ************** 0.074923689 189680 ***************** lookup ************** 0.089702578 335100 ***************** lookup ************** 0.151553955 334685 ***************** lookup ************** 0.117808430 167700 ***************** lookup ************** 0.119441314 336813 ***************** lookup ************** 0.120572137 334684 ***************** lookup ************** 0.124718126 21234 ***************** lookup ************** 0.131124745 84407 ***************** lookup ************** 0.132442945 334696 ***************** lookup ************** 0.140938740 335094 ***************** lookup ************** 0.201637910 167735 ***************** lookup ************** 0.164059859 334687 ***************** lookup ************** 0.252930745 334695 ***************** lookup ************** 0.278037098 341818 0.291815990 ********* Unfinished IO: buffer/disk 50000015000 3:439888512^\scratch_metadata_5 341818 0.291822084 MSG FSnd: nsdMsgReadExt msg_id 2894690129 Sduration 199.688 + us 100041 0.025061905 MSG FRep: nsdMsgReadExt msg_id 1012644746 Rduration 266959.867 us Rlen 0 Hduration 266963.954 + us Elapsed trace time: 0.292021772 seconds Elapsed trace time from first VFS call to last: 0.292021771 Time idle between VFS calls: 0.001436519 seconds Operations stats: total time(s) count avg-usecs wait-time(s) avg-usecs rdwr 0.000831801 4 207.950 read_inode2 0.000082347 31 2.656 pagein 0.000033905 3 11.302 revalidate 0.000013109 156 0.084 open 0.000237969 22 10.817 lookup 1.233407280 10 123340.728 delete_inode 0.000013877 33 0.421 permission 0.000046486 8 5.811 release 0.000172456 21 8.212 mmap 0.000064411 2 32.206 llseek 0.000000391 2 0.196 readdir 0.000213657 36 5.935 User thread stats: GPFS-time(sec) Appl-time GPFS-% Appl-% Ops 335094 0.053506265 0.000170270 99.68% 0.32% 16 167700 0.000008522 0.000027547 23.63% 76.37% 2 167776 0.000008293 0.000019462 29.88% 70.12% 2 334684 0.000023562 0.000160872 12.78% 87.22% 8 349767 0.000000467 0.250029787 0.00% 100.00% 5 84407 0.000000230 0.000017947 1.27% 98.73% 2 334685 0.000028543 0.000094147 23.26% 76.74% 8 406314 0.221755229 0.000009720 100.00% 0.00% 2 334694 0.000024913 0.000125229 16.59% 83.41% 10 335096 0.254359005 0.000240785 99.91% 0.09% 18 334695 0.000028966 0.000127823 18.47% 81.53% 10 334686 0.223770082 0.000267271 99.88% 0.12% 24 334687 0.000031265 0.000132905 19.04% 80.96% 9 334696 0.000033808 0.000131131 20.50% 79.50% 9 129075 0.000000102 0.000000000 100.00% 0.00% 1 341842 0.000000318 0.000000000 100.00% 0.00% 1 335100 0.059518133 0.000287934 99.52% 0.48% 19 224423 0.000000471 0.000000000 100.00% 0.00% 1 336812 0.000042720 0.000193294 18.10% 81.90% 10 21233 0.000556984 0.000083399 86.98% 13.02% 11 289606 0.000000088 0.000018043 0.49% 99.51% 2 362246 0.014440188 0.000046516 99.68% 0.32% 4 21234 0.000524848 0.000162353 76.37% 23.63% 13 336813 0.000046426 0.000175666 20.90% 79.10% 9 3339 0.000011816 0.272396876 0.00% 100.00% 29 341818 0.000000778 0.000000000 100.00% 0.00% 1 167735 0.000007866 0.000049468 13.72% 86.28% 3 175480 0.000000278 0.000000000 100.00% 0.00% 1 336006 0.000001170 0.250020470 0.00% 100.00% 16 44777 0.000000367 0.250149757 0.00% 100.00% 6 189680 0.000002717 0.000006518 29.42% 70.58% 1 184839 0.000003001 0.250144214 0.00% 100.00% 35 145858 0.000000687 0.000000000 100.00% 0.00% 1 333972 0.218656404 0.000043897 99.98% 0.02% 4 334691 0.187695040 0.000295117 99.84% 0.16% 25 # total App-read/write = 7 Average duration = 0.000123672 sec # time(sec) count % %ile read write avgBytesR avgBytesW 0.000500 7 1.000000 1.000000 7 0 1172 0 -----Original Message----- From: gpfsug-discuss-boun...@spectrumscale.org <gpfsug-discuss-boun...@spectrumscale.org> On Behalf Of Uwe Falke Sent: Wednesday, November 11, 2020 8:57 PM To: gpfsug main discussion list <gpfsug-discuss@spectrumscale.org> Subject: Re: [gpfsug-discuss] Poor client performance with high cpu usage of mmfsd process Hi, Kamil, I suppose you'd rather not see such an issue than pursue the ugly work-around to kill off processes. In such situations, the first looks should be for the GPFS log (on the client, on the cluster manager, and maybe on the file system manager) and for the current waiters (that is the list of currently waiting threads) on the hanging client. -> /var/adm/ras/mmfs.log.latest mmdiag --waiters That might give you a first idea what is taking long and which components are involved. Also, mmdiag --iohist shows you the last IOs and some stats (service time, size) for them. Either that clue is already sufficient, or you go on (if you see DIO somewhere, direct IO is used which might slow down things, for example). GPFS has a nice tracing which you can configure or just run the default trace. Running a dedicated (low-level) io trace can be achieved by mmtracectl --start --trace=io --tracedev-write-mode=overwrite -N <your_critical_node> then, when the issue is seen, stop the trace by mmtracectl --stop -N <your_critical_node> Do not wait to stop the trace once you've seen the issue, the trace file cyclically overwrites its output. If the issue lasts some time you could also start the trace while you see it, run the trace for say 20 secs and stop again. On stopping the trace, the output gets converted into an ASCII trace file named trcrpt.*(usually in /tmp/mmfs, check the command output). There you should see lines with FIO which carry the inode of the related file after the "tag" keyword. example: 0.000745100 25123 TRACE_IO: FIO: read data tag 248415 43466 ioVecSize 8 1st buf 0x299E89BC000 disk 8D0 da 154:2083875440 nSectors 128 err 0 finishTime 1563473283.135212150 -> inode is 248415 there is a utility , tsfindinode, to translate that into the file path. you need to build this first if not yet done: cd /usr/lpp/mmfs/samples/util ; make , then run ./tsfindinode -i <inode_num> <fs_mount_point> For the IO trace analysis there is an older tool : /usr/lpp/mmfs/samples/debugtools/trsum.awk. Then there is some new stuff I've not yet used in /usr/lpp/mmfs/samples/traceanz/ (always check the README) Hope that halps a bit. Mit freundlichen Grüßen / Kind regards Dr. Uwe Falke IT Specialist Hybrid Cloud Infrastructure / Technology Consulting & Implementation Services +49 175 575 2877 Mobile Rathausstr. 7, 09111 Chemnitz, Germany uwefa...@de.ibm.com IBM Services IBM Data Privacy Statement IBM Deutschland Business & Technology Services GmbH Geschäftsführung: Sven Schooss, Stefan Hierl Sitz der Gesellschaft: Ehningen Registergericht: Amtsgericht Stuttgart, HRB 17122 From: "Czauz, Kamil" <kamil.cz...@squarepoint-capital.com> To: "gpfsug-discuss@spectrumscale.org" <gpfsug-discuss@spectrumscale.org> Date: 11/11/2020 23:36 Subject: [EXTERNAL] [gpfsug-discuss] Poor client performance with high cpu usage of mmfsd process Sent by: gpfsug-discuss-boun...@spectrumscale.org We regularly run into performance issues on our clients where the client seems to hang when accessing any gpfs mount, even something simple like a ls could take a few minutes to complete. This affects every gpfs mount on the client, but other clients are working just fine. Also the mmfsd process at this point is spinning at something like 300-500% cpu. The only way I have found to solve this is by killing processes that may be doing heavy i/o to the gpfs mounts - but this is more of an art than a science. I often end up killing many processes before finding the offending one. My question is really about finding the offending process easier. Is there something similar to iotop or a trace that I can enable that can tell me what files/processes and being heavily used by the mmfsd process on the client? -Kamil Confidentiality Note: This e-mail and any attachments are confidential and may be protected by legal privilege. If you are not the intended recipient, be aware that any disclosure, copying, distribution or use of this e-mail or any attachment is prohibited. If you have received this e-mail in error, please notify us immediately by returning it to the sender and delete this copy from your system. We will use any personal information you give to us in accordance with our Privacy Policy which can be found in the Data Protection section on our corporate website http://www.squarepoint-capital.com. Please note that e-mails may be monitored for regulatory and compliance purposes. Thank you for your cooperation. _______________________________________________ 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 Confidentiality Note: This e-mail and any attachments are confidential and may be protected by legal privilege. If you are not the intended recipient, be aware that any disclosure, copying, distribution or use of this e-mail or any attachment is prohibited. If you have received this e-mail in error, please notify us immediately by returning it to the sender and delete this copy from your system. We will use any personal information you give to us in accordance with our Privacy Policy which can be found in the Data Protection section on our corporate website www.squarepoint-capital.com. Please note that e-mails may be monitored for regulatory and compliance purposes. Thank you for your cooperation. _______________________________________________ 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