Description of problem:
Everytime when we tried to create a database on an ext3 filesystem we got
the "kernel: ENOMEM in journal_alloc_journal_head,retrying." error and the
machine rebooted itself.
The error could be prevented if we mounted the filesystem as an ext2 before
creating the database.
We're using the with AS2.1 supplied cluster.
Version-Release number of selected component (if applicable):
Sybase Adaptive Server Enterprise/126.96.36.199/EBF 10431 ESD#2/P/
Linux Intel/Linux 2.2.14-5.0smp i686/rel12501/1792/32-bit/OPT/
Mon Jul 29 22:24:47 2002
Create Database on ext3 filesystem
Steps to Reproduce:
The following commands were executed:
/* Generated on 20030326162320 by user XXXXXX*/
create database XXXXXXX on XXXXXXXX = 2000 log on XXXXXXXX = 512
/* Generated on 20030325221626 by user XXXXXX */
disk init name = XXXXXXXX,
vstart = 0,
physname = '/XXX/XXXXXXXX/XXXX/XXXXXX.XXXXXXXX',
size = 1048576,
vdevno = 7
(sensitive Information where replaced by a 'X')
The machines produces the following errors and reboots:
Mar 26 14:26:31 XXXXXXXX kernel: ENOMEM in journal_alloc_journal_head,
Mar 26 14:26:32 XXXXXXXX portmap: connect from 172.19.232.79 to
callit(ypserv): request from unauthorized host
Mar 26 14:26:37 XXXXXXXX kernel: ENOMEM in journal_alloc_journal_head,
To prevent this error:
- Stop Database
- remount filesystems with ext2
- create database
- restart cluster service
Create database on ext3 without reboot
A Service Request is already opened for this:
'Service Request: 230822'
This problem was addressed in the latest kernel errata(e.24).
As it's been about 3 months since this was said to have been fixed, with no
reply saying otherwise, I'm closing this bug.