Red Hat Bugzilla – Bug 102786
ENOMEM in journal_alloc_journal_head, retrying - system hangs
Last modified: 2007-04-18 12:57:01 EDT
From Bugzilla Helper:
User-Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; T312461)
Description of problem:
This is becoming a regular occurance for this server ( 4 times in the past
three weeks )
Receive the following over and over again on the console, system console will
not accept any keyboard input.
ENOMEM in journal_alloc_journal_head, retrying.
ENOMEM in do_get_write_access, retrying
ENOMEM in journal_get_undo_access_Rsmp_10be8bf9, retrying
Version-Release number of selected component (if applicable):
Steps to Reproduce:
I am unsure of the cause and have tried cp large amounts of data as per bug
84738 but nothing.
Dell 6650, 12gigs of ram, broadcom bc5700 driver version 6.0.5
19 Oracle instances connected to a Hitachi San using an emulex driver.
[root@gomez root]# uname -a
Linux gomez 2.4.18-26 #12 SMP Thu May 8 18:19:40 EST 2003 i686 unknown
Those warnings simply tell you that ext3 noticed a serious shortage of available
kernel memory. ext3 was retrying the allocations; other places in the VFS do
the same thing, but only ext3 emits these warnings. So, there's no telling what
is actually wedged: under severe memory pressure, you _expect_ to see warnings
only from ext3, because that's the only component that bothers to tell you that
We'd need capture of at least alt-sysrq-t and alt-sysrq-p output to take this
much further. Serial console would help to capture that.
No activity in 4 months, pending EOL.. closing.