LINUX.ORG.RU
ФорумAdmin

bonding не запускается.. ругается страшным матом


0

0

бондинг мод 0 для увеличения пропускной способности сети. настроил бондинг на лольном тестовом сервере. Всё работает. настроил на боевом. получаю жуткие тормоза. При этом характерно что после ifdown eth0 and ifup bond0 сервер некоторое время (более минуты) не отвечает на запросы совсем. Затем отвечает с жуткими тормозами. После ifdown bond0 and ifup eth0 та же исмтория. молчвние около минуты, а затем нормальная работа. "зивисания" гораздо дольше времени выполнения ifup ifdown.. при ifdown eth0 ifup bond0 dmesg | less матерится так:

★★★★★

ADDRCONF(NETDEV_UP): eth0: link is not ready
bonding: bond0: enslaving eth0 as an active interface with a down link.
eth2: -- ERROR --
Class: internal Software error
Nr: 0x2bd
Msg: TWSI: transfer does not complete
ADDRCONF(NETDEV_UP): eth2: link is not ready
bonding: bond0: enslaving eth2 as an active interface with a down link.
tg3: eth0: Link is up at 100 Mbps, full duplex.
tg3: eth0: Flow control is off for TX and off for RX.
ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
bonding: bond0: link status definitely up for interface eth0.
eth2: network connection up using port A
speed: 100
autonegotiation: yes
duplex mode: full
flowctrl: none
irq moderation: disabled
scatter-gather: disabled
tx-checksum: disabled
rx-checksum: disabled
bonding: bond0: link status definitely up for interface eth2.
ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready
bond0: no IPv6 routers present
eth0: no IPv6 routers present
eth2: no IPv6 routers present

eth2: network connection down
bonding: bond0: released all slaves
Ethernet Channel Bonding Driver: v3.0.1 (January 9, 2006)
bonding: In ALB mode you might experience client disconnections upon reconnection of a link if the bonding module updelay parameter (0 msec
) is incompatible with the forwarding delay time of the switch
bonding: MII link monitoring set to 100 ms
ADDRCONF(NETDEV_UP): eth0: link is not ready
bonding: bond0: enslaving eth0 as an active interface with a down link.
eth2: -- ERROR --
Class: internal Software error
Nr: 0x2bd
Msg: TWSI: transfer does not complete
ADDRCONF(NETDEV_UP): eth2: link is not ready
bonding: bond0: Error: dev_set_mac_address of dev eth2 failed! ALB mode requires that the base driver support setting the hw address also w
hen the network device's interface is open
eth2: network connection down
eth2: -- ERROR --
Class: internal Software error
Nr: 0x2bd
Msg: TWSI: transfer does not complete
ADDRCONF(NETDEV_UP): eth2: link is not ready
bonding: bond0: Error: dev_set_mac_address of dev eth2 failed! ALB mode requires that the base driver support setting the hw address also w
hen the network device's interface is open
eth2: network connection down
tg3: eth0: Link is up at 100 Mbps, full duplex.
tg3: eth0: Flow control is off for TX and off for RX.
ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
bonding: bond0: link status definitely up for interface eth0.
bonding: bond0: making interface eth0 the new active one.
Debug: sleeping function called from invalid context at mm/slab.c:2731
in_atomic():1, irqs_disabled():0

Call Trace: <IRQ> <ffffffff80176c94>{kmem_cache_alloc+32}
<ffffffff8027296d>{__alloc_skb+56} <ffffffff8027e593>{rtmsg_ifinfo+35}
<ffffffff8027e65c>{rtnetlink_event+64} <ffffffff802d2c1c>{notifier_call_chain+28}
<ffffffff802756fb>{dev_set_mac_address+83} <ffffffff883727a9>{:bond0:alb_set_slave_mac_addr+65}
<ffffffff88372bb5>{:bond0:alb_swap_mac_addr+145} <ffffffff8836df10>{:bond0:bond_change_active_slave+475}
<ffffffff8836f8f3>{:bond0:bond_mii_monitor+916} <ffffffff8836f55f>{:bond0:bond_mii_monitor+0}
<ffffffff80139ba4>{run_timer_softirq+348} <ffffffff8013596e>{__do_softirq+85}
<ffffffff8010b9c2>{call_softirq+30} <ffffffff8010c920>{do_softirq+44}
<ffffffff801097b4>{default_idle+0} <ffffffff8010b366>{apic_timer_interrupt+98} <EOI>
<ffffffff801097df>{default_idle+43} <ffffffff8010989f>{cpu_idle+151}
<ffffffff8011823b>{start_secondary+1240}
RTNL: assertion failed at net/ipv4/devinet.c (985)

Call Trace: <IRQ> <ffffffff802b5dfc>{inetdev_event+95}
<ffffffff8028f6d5>{rt_run_flush+126} <ffffffff802d2c1c>{notifier_call_chain+28}
<ffffffff802756fb>{dev_set_mac_address+83} <ffffffff883727a9>{:bond0:alb_set_slave_mac_addr+65}
<ffffffff88372bb5>{:bond0:alb_swap_mac_addr+145} <ffffffff8836df10>{:bond0:bond_change_active_slave+475}
<ffffffff8836f8f3>{:bond0:bond_mii_monitor+916} <ffffffff8836f55f>{:bond0:bond_mii_monitor+0}
<ffffffff80139ba4>{run_timer_softirq+348} <ffffffff8013596e>{__do_softirq+85}
<ffffffff8010b9c2>{call_softirq+30} <ffffffff8010c920>{do_softirq+44}
<ffffffff801097b4>{default_idle+0} <ffffffff8010b366>{apic_timer_interrupt+98} <EOI>
<ffffffff801097df>{default_idle+43} <ffffffff8010989f>{cpu_idle+151}
<ffffffff8011823b>{start_secondary+1240}
bond0: no IPv6 routers present
eth0: no IPv6 routers present

bonding: bond0: released all slaves
Ethernet Channel Bonding Driver: v3.0.1 (January 9, 2006)
bonding: MII link monitoring set to 100 ms
ADDRCONF(NETDEV_UP): eth0: link is not ready
bonding: bond0: enslaving eth0 as an active interface with a down link.
eth2: -- ERROR --
Class: internal Software error
Nr: 0x2bd
Msg: TWSI: transfer does not complete
ADDRCONF(NETDEV_UP): eth2: link is not ready
bonding: bond0: enslaving eth2 as an active interface with a down link.
tg3: eth0: Link is up at 100 Mbps, full duplex.
tg3: eth0: Flow control is off for TX and off for RX.
ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
bonding: bond0: link status definitely up for interface eth0.
eth2: network connection up using port A
speed: 100
autonegotiation: yes
duplex mode: full
flowctrl: none
irq moderation: disabled
scatter-gather: disabled
tx-checksum: disabled
rx-checksum: disabled
bonding: bond0: link status definitely up for interface eth2.
ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready
bond0: no IPv6 routers present
eth0: no IPv6 routers present
eth2: no IPv6 routers present

eth2: network connection down
bonding: bond0: released all slaves
ADDRCONF(NETDEV_UP): eth0: link is not ready
tg3: eth0: Link is up at 100 Mbps, full duplex.
tg3: eth0: Flow control is off for TX and off for RX.
ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
eth2: -- ERROR --
Class: internal Software error
Nr: 0x2bd
Msg: TWSI: transfer does not complete
ADDRCONF(NETDEV_UP): eth2: link is not ready
eth2: network connection up using port A
speed: 100
autonegotiation: yes
duplex mode: full
flowctrl: none
irq moderation: disabled
scatter-gather: disabled
tx-checksum: disabled
rx-checksum: disabled
ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready
eth2: no IPv6 routers present

AndreyKl ★★★★★
() автор топика
Ответ на: комментарий от AndreyKl

а в /var/log/messages нахожу следующее:


May 29 18:11:38 linux66 ifdown: eth0 device: Broadcom Corporation NetXtreme BCM5704 Gigabit Ethernet (rev 10)
May 29 18:11:38 linux66 ifdown: eth0 configuration: eth-id-00:30:48:58:e8:5e
May 29 18:11:44 linux66 ifdown: eth2 device: Linksys Gigabit Network Adapter (rev 12)
May 29 18:11:44 linux66 ifdown: eth2 configuration: eth-id-00:18:f8:08:07:89
May 29 18:11:49 linux66 ifup: bond0
May 29 18:11:49 linux66 kernel: Ethernet Channel Bonding Driver: v3.0.1 (January 9, 2006)
May 29 18:11:49 linux66 kernel: bonding: MII link monitoring set to 100 ms
May 29 18:11:49 linux66 ifup: bond0 enslaving interfaces: eth0 eth2
May 29 18:11:49 linux66 kernel: ADDRCONF(NETDEV_UP): eth0: link is not ready
May 29 18:11:49 linux66 kernel: bonding: bond0: enslaving eth0 as an active interface with a down link.
May 29 18:11:50 linux66 kernel: eth2: -- ERROR --
May 29 18:11:50 linux66 kernel: Class: internal Software error
May 29 18:11:50 linux66 kernel: Nr: 0x2bd
May 29 18:11:50 linux66 kernel: Msg: TWSI: transfer does not complete
May 29 18:11:50 linux66 kernel: ADDRCONF(NETDEV_UP): eth2: link is not ready
May 29 18:11:50 linux66 kernel: bonding: bond0: enslaving eth2 as an active interface with a down link.
May 29 18:11:51 linux66 kernel: tg3: eth0: Link is up at 100 Mbps, full duplex.
May 29 18:11:51 linux66 kernel: tg3: eth0: Flow control is off for TX and off for RX.
May 29 18:11:51 linux66 kernel: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
May 29 18:11:51 linux66 kernel: bonding: bond0: link status definitely up for interface eth0.
May 29 18:11:51 linux66 kernel: eth2: network connection up using port A
May 29 18:11:51 linux66 kernel: speed: 100
May 29 18:11:51 linux66 kernel: autonegotiation: yes
May 29 18:11:51 linux66 kernel: duplex mode: full
May 29 18:11:51 linux66 kernel: flowctrl: none
May 29 18:11:51 linux66 kernel: irq moderation: disabled
May 29 18:11:51 linux66 kernel: scatter-gather: disabled
May 29 18:11:51 linux66 kernel: tx-checksum: disabled
May 29 18:11:51 linux66 kernel: rx-checksum: disabled
May 29 18:11:51 linux66 kernel: bonding: bond0: link status definitely up for interface eth2.
May 29 18:11:52 linux66 kernel: ADDRCONF(NETDEV_CHANGE): eth2: link becomes ready
May 29 18:12:00 linux66 kernel: bond0: no IPv6 routers present
May 29 18:12:02 linux66 kernel: eth0: no IPv6 routers present
May 29 18:12:02 linux66 kernel: eth2: no IPv6 routers present


есть ли идеи, как лечить?

или может другой способ выполнить задачу(задействовать вторую сетевуху для расширения канала)?

AndreyKl ★★★★★
() автор топика
Ответ на: комментарий от anonymous

по отдельности работают. Всего три интерфейса, два для бондинга, и один в локалку смотрит (по нему я захожу на машину). При перезапуске бондинга локалка не отваливается, значит, не перепутал.

AndreyKl ★★★★★
() автор топика
Ответ на: комментарий от AndreyKl

Большая просьба найдете причину, напишите сюда. Я вчера наступил на эти же грабли вот сижу с задумчивым взглядом. Пока решения не нашел.

pylon
()
Ответ на: комментарий от pylon

Одна интересная особенность, если дать пинг с этой (бондовой) машины все работает. Если пинговать бондовую машину иногда получается провал, достаточно длинный, потом пробивает. Я уж тут подумал, может проблема в свиче. Кстати в логах у меня чисто, говорит что все ОК.

pylon
()
Ответ на: комментарий от pylon

пока ничего.. на локалке не воспроизводится, возможных причины вижу две:

1) проблема с драйвером сетевухи
2) проблема со свичами.

решил для начала обновить ядро на сервере.
сейчас 2.6.16.13-4-smp
буду ставить 2.6.16.27.
О результатах отпишусь.

ЗЫ. Сетевухи:

Super Micro NetXtreme BCM5704 Gigabit Ethernet и
Linksys EG1064 v2 Instant Gigabit Network Adapter

AndreyKl ★★★★★
() автор топика
Вы не можете добавлять комментарии в эту тему. Тема перемещена в архив.