> On Feb 18, 2016, at 11:31 AM, Dan Langille <d...@langille.org> wrote: > >> On Feb 18, 2016, at 6:44 AM, Josh Fisher <jfis...@pvct.com >> <mailto:jfis...@pvct.com>> wrote: >> >> On 2/17/2016 6:07 PM, Dan Langille wrote: >>>> On Feb 17, 2016, at 9:57 AM, Martin Simmons < >>>> <mailto:mar...@lispworks.com>mar...@lispworks.com >>>> <mailto:mar...@lispworks.com>> wrote: >>>> >>>>>>>>> On Tue, 16 Feb 2016 18:59:10 -0500, Dan Langille said: >>>>> >>>>> As a test, I scp'd over all the volumes for one job and timed it: >>>>> >>>>> real 167m15.956s >>>>> user 117m51.006s >>>>> sys 23m56.653s >>>>> >>>>> FWIW, these copies were done while a ZFS scrub was underway, so if >>>>> anything, the potential throughput is higher. >>>>> >>>>> The job above, took nearly 11 hours. Something is up. >>>>> >>>>> I tried another test run, just now, but cancelled it after the first two >>>>> Volumes were read; it was going at about the same rate as the full job >>>>> mentioned above. >>>>> >>>>> What might account for this vast difference? >>>> >>>> You could try profiling it, e.g >>>> http://www.brendangregg.com/FlameGraphs/cpuflamegraphs.html#DTrace >>>> <http://www.brendangregg.com/FlameGraphs/cpuflamegraphs.html#DTrace> >>> >>> >>> This is a great idea. Have you done this before? I haven't and my first >>> attempt seems to be interesting. >>> >>> http://www.langille.org/tmp/bacula-copy.svg >>> <http://www.langille.org/tmp/bacula-copy.svg> >>> >>> It didn't run very long, and thus I'm not sure how relevant the information >>> is >> >> Interesting, Dan. I wonder what it looks like for that same job with >> encryption turned off. If I am reading the graph correctly, it looks like >> AES encryption is more than doubling the CPU usage. > > To be clear, this was TLS. I just disabled TLS for bacula-sd and ran it > again: > > http://www.langille.org/tmp/bacula-copy-without-tls.svg > <http://www.langille.org/tmp/bacula-copy-without-tls.svg> > > I don't know how representative these brief captures are.
... but here is a second dtrace during the same job: http://www.langille.org/tmp/bacula-copy-without-tls2.svg <http://www.langille.org/tmp/bacula-copy-without-tls2.svg> I am happy to provide more dtrace results: # dtrace -x ustackframes=100 -n 'profile-99 /execname == "bacula-sd" && arg1/ {@[ustack()] = count(); } tick-60s { exit(0); }' -o out.stacks2 dtrace: description 'profile-99 ' matched 2 probes -- Dan Langille - BSDCan / PGCon d...@langille.org
signature.asc
Description: Message signed with OpenPGP using GPGMail
------------------------------------------------------------------------------ Site24x7 APM Insight: Get Deep Visibility into Application Performance APM + Mobile APM + RUM: Monitor 3 App instances at just $35/Month Monitor end-to-end web transactions and take corrective actions now Troubleshoot faster and improve end-user experience. Signup Now! http://pubads.g.doubleclick.net/gampad/clk?id=272487151&iu=/4140
_______________________________________________ Bacula-users mailing list Bacula-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/bacula-users
