Date: Fri, 11 Sep 2015 10:19:10 -0700 From: hiren panchasara <hiren@strugglingcoder.info> To: Dmitry Sivachenko <trtrmitya@gmail.com> Cc: FreeBSD Net <freebsd-net@freebsd.org> Subject: Re: netisr process eats 100% cpu Message-ID: <20150911171910.GN64965@strugglingcoder.info> In-Reply-To: <EA9AD61B-E709-4387-8D5E-A33E25475D81@gmail.com> References: <C4039E45-3DC7-4E1A-9E46-853FE812B906@gmail.com> <EA9AD61B-E709-4387-8D5E-A33E25475D81@gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
--IpgPcFyQO6wM49Um Content-Type: text/plain; charset=us-ascii Content-Disposition: inline Content-Transfer-Encoding: quoted-printable On 09/11/15 at 12:46P, Dmitry Sivachenko wrote: >=20 > > hiren panchasara <hiren at strugglingcoder.info> wrote: > >=20 > > Unsure at the moment if loopback is causing the trouble for you or not. > > See: >=20 > (please keep me CC'ed, I am not subscribed to -net) >=20 >=20 > > > > https://lists.freebsd.org/pipermail/freebsd-net/2015-February/041239.ht= ml > > > > >=20 > Yes, this thread looks similar. >=20 >=20 > > You may want to try: > > 1) pmcstat and see if you can catch something >=20 > What particular should I look for? Here is first lines of pmcstat -T -S = instructions -w 1: > PMC: [INSTR_RETIRED_ANY] Samples: 157198 (100.0%) , 0 unresolved >=20 > %SAMP IMAGE FUNCTION CALLERS > 13.2 kernel cpu_search_highest cpu_search_highest:12.0 sched_idlet= d:1.2 > 8.3 kernel ipfw_chk ipfw_check_packet > 3.1 myprogram memsetAVX _ZN12TLz4Compress7DoWriteEPKv > 2.3 kernel tcp_output tcp_usr_send:1.0 tcp_do_segment:0.9 >=20 >=20 > > 2) disable checksum on localhost >=20 >=20 > I tried, but nothing has changed. >=20 >=20 > > 3) look at netisr settings. sysctl net.isr o/p and how it looks under > > netstat -Q. I am not sure if adding more threads to netisr via >=20 >=20 > What should I look for? >=20 >=20 > > net.isr.numthreads would help. (Note its a loader.conf variable) >=20 >=20 > This netisr load looks parasitical to me (as I noted, moving haproxy to a= separate machine does not burn CPU cycles on netisr, why is localhost spec= ial?) >=20 > Even if adding more threads to netisr would boost network utilization, wo= uldn't those CPU cycles spent for netisr just be a waste of energy? I have= other tasks for these CPU. > I am not sure what keep cpu busy with netisr when localhost is involved. You may want to post o/p of=20 # sysctl net.isr # netstat -Q # sysctl net.inet | grep queue A suggestion I see at https://calomel.org/freebsd_network_tuning.html is to increase localhost n/w buffers. Not sure if this'll help your case. net.local.stream.sendspace=3D164240 # (default 8192) net.local.stream.recvspace=3D164240 # (default 8192) Now I'll let someone else with more ideas/clues comment. Cheers, Hiren --IpgPcFyQO6wM49Um Content-Type: application/pgp-signature -----BEGIN PGP SIGNATURE----- Version: GnuPG v2 iQF8BAABCgBmBQJV8w0KXxSAAAAAAC4AKGlzc3Vlci1mcHJAbm90YXRpb25zLm9w ZW5wZ3AuZmlmdGhob3JzZW1hbi5uZXRBNEUyMEZBMUQ4Nzg4RjNGMTdFNjZGMDI4 QjkyNTBFMTU2M0VERkU1AAoJEIuSUOFWPt/l0m8H/0xmKSS0SNp9uxn9I2k39uNG 7Mx+w2S4FlAvzOuVEhK8kteWSWTdaIhLIAYj45edU4tJNB0oBGmkvY7rRWqmwydf +2QaedAw1eph6N3NkcsVEgcx6muSXcSGYG5DgX7Tfrz2UKuf3qiXrJfSw6JHmqTO tTycvuAjNrEc4OLnm7R8OWsk+hwIHkwc/+K7AEUMCr+pnJ5DBt+yMIhIEDoLmqWV 8S+AMHe7zYJKHno45csqXwZuSNSSCcv3YhXmPFqcgtDzNIQt1Xo5Q1UPxF0XSGEa ypQG1H9m1QtaT6/lfefYWDk75zBK6dbM9yBx6QauN1ij6dAx21obJzm8kwgneVM= =3bgx -----END PGP SIGNATURE----- --IpgPcFyQO6wM49Um--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20150911171910.GN64965>