Bug 761442

Latest response

Tenemos instalado redhat EL 6.5 en producción, de 64 bits. En ella tenemos instalado Oracle y nos aparece este error:

Jun 22 23:52:13 bd01 kernel: oracle: page allocation failure. order:4, mode:0x20
Jun 22 23:52:13 bd01 kernel: Pid: 35754, comm: oracle Not tainted 2.6.32-431.17.1.el6.x86_64 #1
Jun 22 23:52:13 bd01 kernel: Call Trace:
Jun 22 23:52:13 bd01 kernel: [] ? __alloc_pages_nodemask+0x74a/0x8d0
Jun 22 23:52:13 bd01 kernel: [] ? kmem_getpages+0x62/0x170
Jun 22 23:52:13 bd01 kernel: [] ? fallback_alloc+0x1ba/0x270
Jun 22 23:52:13 bd01 kernel: [] ? ____cache_alloc_node+0x99/0x160
Jun 22 23:52:13 bd01 kernel: [] ? kmem_cache_alloc_node_trace+0x90/0x200
Jun 22 23:52:13 bd01 kernel: [] ? __kmalloc_node+0x4d/0x60
Jun 22 23:52:13 bd01 kernel: [] ? __alloc_skb+0x7a/0x180
Jun 22 23:52:13 bd01 kernel: [] ? skb_copy+0x40/0xb0
Jun 22 23:52:13 bd01 kernel: [] ? tg3_start_xmit+0xa8c/0xd80 [tg3]
Jun 22 23:52:13 bd01 kernel: [] ? dev_hard_start_xmit+0x224/0x480
Jun 22 23:52:13 bd01 kernel: [] ? sch_direct_xmit+0x15a/0x1c0
Jun 22 23:52:13 bd01 kernel: [] ? dev_queue_xmit+0x228/0x320
Jun 22 23:52:13 bd01 kernel: [] ? ip_finish_output+0x148/0x310
Jun 22 23:52:13 bd01 kernel: [] ? ip_output+0xb8/0xc0
Jun 22 23:52:13 bd01 kernel: [] ? ip_local_out+0x25/0x30
Jun 22 23:52:13 bd01 kernel: [] ? ip_queue_xmit+0x190/0x420
Jun 22 23:52:13 bd01 kernel: [] ? __alloc_pages_nodemask+0x113/0x8d0
Jun 22 23:52:13 bd01 kernel: [] ? tcp_transmit_skb+0x40e/0x7b0
Jun 22 23:52:13 bd01 kernel: [] ? tcp_write_xmit+0x230/0xa90
Jun 22 23:52:13 bd01 kernel: [] ? __kmalloc_node+0x4d/0x60
Jun 22 23:52:13 bd01 kernel: [] ? __alloc_skb+0x7a/0x180
Jun 22 23:52:13 bd01 kernel: [] ? tcp_push_one+0x30/0x40
Jun 22 23:52:13 bd01 kernel: [] ? tcp_sendmsg+0x9cc/0xa20
Jun 22 23:52:13 bd01 kernel: [] ? sock_sendmsg+0x123/0x150
Jun 22 23:52:13 bd01 kernel: [] ? autoremove_wake_function+0x0/0x40
Jun 22 23:52:13 bd01 kernel: [] ? futex_wake+0x10e/0x120
Jun 22 23:52:13 bd01 kernel: [] ? do_futex+0x100/0xb60
Jun 22 23:52:13 bd01 kernel: [] ? sys_sendto+0x139/0x190
Jun 22 23:52:13 bd01 kernel: [] ? thread_return+0x4e/0x760
Jun 22 23:52:13 bd01 kernel: [] ? sys_futex+0x7b/0x170
Jun 22 23:52:13 bd01 kernel: [] ? audit_syscall_entry+0x1d7/0x200
Jun 22 23:52:13 bd01 kernel: [] ? system_call_fastpath+0x16/0x1b
Jun 23 00:35:33 bd01 kernel: swapper: page allocation failure. order:4, mode:0x20

MD
Log in to join the conversation

Responses

rhn-support-mflitter's picture Red Hat Guru 1638 points
23 June 2014 1:29 PM

Hi Felix,

Have you tried the sysctl shown in the bugzilla?

sysctl -w vm.zone_reclaim_mode=1

If this does not work, are you in a position to capture a vmcore open a new support ticket with the details?

Many thanks,
Mark

MD Community Member 25 points
23 June 2014 2:12 PM

Thank's. I had sawn this comment on internet. The problem that I have, It is on a Production enviornment, this key is secure?

MD Community Member 25 points
23 June 2014 2:14 PM

Thank's. I had seen this comment on internet. The problem that I have, It is on a Production enviornment, this key is secure?

rhn-support-mflitter's picture Red Hat Guru 1638 points
23 June 2014 2:33 PM

Hi Felix,

Unfortunately every change comes with an associated risk that can only be evaluated by the system admin of that server.

Under some workloads, you may find that you take a performance hit by setting vm.zone_reclaim_mode=1, but this can only be determined through testing.

I would recommend that you try this in your dev/test environment first, or schedule downtime and set the parameter in /etc/sysctl.conf to be picked up on next reboot, to avoid making a change on a live server.

In addition, it should be noted that BZ 761442 is still open and that this should be considered a workaround until such time that the root cause is found.

Best regards,
Mark

Close

Welcome! Check out the Getting Started with Red Hat page for quick tours and guides for common tasks.