tez mam ten problem i aby nie przedluzac wkleję wszystko co przedmowca
maszyna już leciwa - 3 lata 2x quad Intel(R) Xeon(R) CPU E5620 @ 2.40GHz z HT przerwania na IBMe z sieciowek nie rozkładały sie samoczynnie po procesorach wiec musiałem troche im pomoc być może upgrade firmware by pomógł , ale od nowości tego nie robiłem.
# cat /proc/interrupts | grep eth 62: 3120224563 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth0-0 63: 1 0 601283296 0 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth0-1 64: 1 0 0 0 354445007 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth0-2 65: 1 0 0 0 0 0 55878171 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth0-3 66: 1 0 0 0 0 0 0 0 4271877651 0 0 0 0 0 0 0 PCI-MSI-edge eth0-4 67: 1 0 0 0 0 0 0 0 0 0 4039073456 0 0 0 0 0 PCI-MSI-edge eth0-5 68: 1 0 0 0 0 0 0 0 0 0 0 0 4172376723 0 0 0 PCI-MSI-edge eth0-6 69: 1 0 0 0 0 0 0 0 0 0 0 0 0 0 4205195798 0 PCI-MSI-edge eth0-7 71: 2 3091601105 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth1-0 72: 1 0 0 394418145 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth1-1 73: 1 0 0 0 0 483541645 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth1-2 74: 1 0 0 0 0 0 0 156099257 0 0 0 0 0 0 0 0 PCI-MSI-edge eth1-3 75: 1 0 0 0 0 0 0 0 0 4237235842 0 0 0 0 0 0 PCI-MSI-edge eth1-4 76: 1 0 0 0 0 0 0 0 0 0 0 4048575793 0 0 0 0 PCI-MSI-edge eth1-5 77: 1 0 0 0 0 0 0 0 0 0 0 0 0 8318720 0 0 PCI-MSI-edge eth1-6 78: 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4168411596 PCI-MSI-edge eth1-7 80: 3130427837 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth2-0 81: 1 0 424933849 0 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth2-1 82: 1 0 0 0 551239842 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth2-2 83: 1 0 0 0 0 0 89592088 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth2-3 84: 1 0 0 0 0 0 0 0 4205109090 0 0 0 0 0 0 0 PCI-MSI-edge eth2-4 85: 1 0 0 0 0 0 0 0 0 0 4156524559 0 0 0 0 0 PCI-MSI-edge eth2-5 86: 1 0 0 0 0 0 0 0 0 0 0 0 4283467891 0 0 0 PCI-MSI-edge eth2-6 87: 1 0 0 0 0 0 0 0 0 0 0 0 0 0 129587482 0 PCI-MSI-edge eth2-7 89: 2 3096616488 0 0 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth3-0 90: 1 0 0 602130018 0 0 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth3-1 91: 1 0 0 0 0 339631942 0 0 0 0 0 0 0 0 0 0 PCI-MSI-edge eth3-2 92: 1 0 0 0 0 0 0 115070982 0 0 0 0 0 0 0 0 PCI-MSI-edge eth3-3 93: 1 0 0 0 0 0 0 0 0 15143274 0 0 0 0 0 0 PCI-MSI-edge eth3-4 94: 1 0 0 0 0 0 0 0 0 0 0 4085039711 0 0 0 0 PCI-MSI-edge eth3-5 95: 1 0 0 0 0 0 0 0 0 0 0 0 0 4270669344 0 0 PCI-MSI-edge eth3-6 96: 1 0 0 0 0 0 0 0 0 0 0 0 0 0 0 4109460081 PCI-MSI-edge eth3-7
# procinfo Linux 2.6.29.6 (root@midas64) (gcc [can't parse]) #??? 16CPU [ibm1.(none)]
Memory: Total Used Free Shared Buffers Mem: 8142904 5640592 2502312 0 533728 Swap: 5863716 0 5863716
Bootup: Sat Sep 1 17:25:58 2012 Load average: 0.18 0.12 0.09 1/410 28072
user : 4d 15:28:38.30 0.3% page in : 0 nice : 0:00:17.16 0.0% page out: 0 system: 1d 4:51:40.41 0.0% swap in : 0 idle : 1233d 23873:1431690:85899381.96 89.3% swap out: 0 uptime: 86d 7:26:03.08 context :4670349204
na interfejsie do internetu # tc class show dev bond0.398 | wc -l 7388 # tc qdisc show dev bond0.398 | wc -l 5465 # tc filter show dev bond0.398 | wc -l 5466
oznaczam pakiety firewallem # iptables-save | wc -l 40476
sumarycznie na pozostałych tyle samo kolejek
nie uzywam nata i mam wylaczony conntrack
przy obciazeniu 4x 300Mbit (około 1,2Gbit ) procki maja 40% w internecie znalazlem opinie ze linux powinien poradzic sobie 1g ruchu na 1Ghz procka, mam 8x2,4 +HT i jakos przy predkosciach 1/20 robia dropy.
#lsmod Module Size Used by ipt_LOG 6244 2 xt_mark 2368 47 cls_fw 5408 144 sch_sfq 6752 10974 sch_hfsc 16640 48 iptable_mangle 3072 1 xt_MARK 2944 34644 xt_length 1984 8649 xt_tcpudp 3424 26056 iptable_filter 3040 1 ip_tables 18896 2 iptable_mangle,iptable_filter x_tables 20168 6 ipt_LOG,xt_mark,xt_MARK,xt_length,xt_tcpudp,ip_tables 8021q 23968 0 garp 8320 1 8021q stp 2660 1 garp bonding 99832 0 ipv6 288704 61 bonding ...
i najbardziej martiwa mnie dropy przy przeciazeniu maszyny # ifconfig bond0 bond0 Link encap:Ethernet HWaddr e4:1f:13:69:5f:98 inet addr:10.0.0.21 Bcast:10.255.255.255 Mask:255.255.255.255 inet6 addr: fe80::e61f:13ff:fe69:5f98/64 Scope:Link UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1 RX packets:449348185043 errors:817 dropped:137983 overruns:0 frame:817 TX packets:448042695121 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:374106422227811 (340.2 TiB) TX bytes:372941169952076 (339.1 TiB)
kombinowałem juz wiele z tuningiem modulu sieciowego w linux i dropy jak byly tak sa , moze tylko troche mniej
na blisko 90 dniowy uptime w liscie procesow soft irq nie dawał o sobie zbytnio znać # ps axuf USER PID %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND root 2 0.0 0.0 0 0 ? S< Sep01 0:00 [kthreadd] root 3 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/0] root 4 0.0 0.0 0 0 ? S< Sep01 1:54 _ [ksoftirqd/0] root 5 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/1] root 6 0.0 0.0 0 0 ? S< Sep01 0:00 _ [ksoftirqd/1] root 7 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/2] root 8 0.0 0.0 0 0 ? S< Sep01 0:05 _ [ksoftirqd/2] root 9 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/3] root 10 0.0 0.0 0 0 ? S< Sep01 0:05 _ [ksoftirqd/3] root 11 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/4] root 12 0.0 0.0 0 0 ? S< Sep01 0:03 _ [ksoftirqd/4] root 13 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/5] root 14 0.0 0.0 0 0 ? S< Sep01 0:03 _ [ksoftirqd/5] root 15 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/6] root 16 0.0 0.0 0 0 ? S< Sep01 0:02 _ [ksoftirqd/6] root 17 0.0 0.0 0 0 ? S< Sep01 0:00 _ [migration/7] root 18 0.0 0.0 0 0 ? S< Sep01 0:02 _ [ksoftirqd/7] root 19 0.0 0.0 0 0 ? S< Sep01 0:02 _ [migration/8] root 20 0.0 0.0 0 0 ? S< Sep01 0:02 _ [ksoftirqd/8] root 21 0.0 0.0 0 0 ? S< Sep01 1:06 _ [migration/9] root 22 0.0 0.0 0 0 ? S< Sep01 0:03 _ [ksoftirqd/9] root 23 0.0 0.0 0 0 ? S< Sep01 0:06 _ [migration/10] root 24 0.0 0.0 0 0 ? S< Sep01 0:04 _ [ksoftirqd/10] root 25 0.0 0.0 0 0 ? S< Sep01 0:06 _ [migration/11] root 26 0.0 0.0 0 0 ? S< Sep01 0:03 _ [ksoftirqd/11] root 27 0.0 0.0 0 0 ? S< Sep01 0:02 _ [migration/12] root 28 0.0 0.0 0 0 ? S< Sep01 0:02 _ [ksoftirqd/12] root 29 0.0 0.0 0 0 ? S< Sep01 1:18 _ [migration/13] root 30 0.0 0.0 0 0 ? S< Sep01 0:03 _ [ksoftirqd/13] root 31 0.0 0.0 0 0 ? S< Sep01 0:05 _ [migration/14] root 32 0.0 0.0 0 0 ? S< Sep01 0:02 _ [ksoftirqd/14] root 33 0.0 0.0 0 0 ? S< Sep01 0:06 _ [migration/15] root 34 0.0 0.0 0 0 ? S< Sep01 0:02 _ [ksoftirqd/15]
a to z maszynki która jest routerem bgp i nie ma podzialu pasma # ifconfig bond0 bond0 Link encap:Ethernet HWaddr 44:f3:fc:b6:a8:b0 inet addr:10.0.0.5 Bcast:10.255.255.255 Mask:255.255.255.255 inet6 addr: fe80::46f3:fcff:feb6:a8b0/64 Scope:Link UP BROADCAST RUNNING MASTER MULTICAST MTU:1500 Metric:1 RX packets:2025283836477 errors:3377 dropped:85412 overruns:0 frame:3377 TX packets:2024600702864 errors:0 dropped:0 overruns:0 carrier:0 collisions:0 txqueuelen:0 RX bytes:1665833476281313 (1.4 PiB) TX bytes:1665447884758706 (1.4 PiB) # procinfo Linux 3.2.13 (root@hive64) (gcc [can't parse]) #??? 24CPU [ro-2-ibm4.(none)]
Memory: Total Used Free Shared Buffers Mem: 8147540 3231924 4915616 0 314900 Swap: 5863720 0 5863720
Bootup: Wed Jul 4 11:33:15 2012 Load average: 0.10 0.08 0.05 1/295 15932
user : 7d 3:41:39.44 0.2% page in : 0 nice : 0:00:22.41 0.0% page out: 0 system: 8:01:28.81 0.0% swap in : 0 idle : 3313d 71592:4295007:257698067.70 94.8% swap out: 0 uptime: 145d 14:38:53.59 context :1541369906
i tez jej zdazy sie cos dropnac