United States (change)
Shortcuts: Downloads Fedora Red Hat Network
Issue #22 August 2006
Red Hat's customer service and support teams receive technical support questions from users all over the world. Red Hat technicians add the questions and answers to Red Hat Knowledgebase on a daily basis. Access to Red Hat Knowledgebase is free. Every month, Red Hat Magazine offers a preview into the Red Hat Knowledgebase by highlighting some of the most recent entries.
As hard disk space is ever increasing, you may get replacement drives that are significantly larger than the original devices that they replace, so this tip will show how to increase the size of a RAID array using larger partitions to replace smaller partitions in the original RAID array.
We will assume that you have a RAID 5 array using three partitions (/dev/sdb1, /dev/sdc1, and /dev/sdd1) on /dev/md0. These partitions are 1 GiB each, giving you about 2 GiB of usable space. You add new disks and create three partitions (/dev/sde1, /dev/sdf1, and /dev/sdg1) of 5 GiB in size. By the end, you should have about 10 GiB of usable space.
After you have created the partitions and set the partitions type to 0xfd, you can add these devices to the array. They will become hot spares:
mdadm /dev/md0 -a /dev/sde1 /dev/sdf1 /dev/sdg1
Fail the original devices one at a time, ensuring that the array rebuilds after each failed device. DO NOT fail more that one of the original devices without verifying that the array has finished rebuilding. If you fail two devices in a RAID 5 array, you may destroy data!
First, fail and remove the first device, and verify that the array has finished rebuilding:
mdadm /dev/md0 -f /dev/sdb1 -r /dev/sdb1 watch cat /proc/mdstat
Once it has finished rebuilding, fail the second device:
mdadm /dev/md0 -f /dev/sdc1 -r /dev/sdc1 watch cat /proc/mdstat
Once it has finished rebuilding, fail the third device:
mdadm /dev/md0 -f /dev/sdd1 -r /dev/sdd1 watch cat /proc/mdstat
After it has finished rebuilding, you have replaced all of the 1 GiB original devices with the new 5 GiB devices. However, we are not finished yet. We have two problems: the RAID array is still only using 1 GiB of my 5 GiB devices, and the filesystem is still 2 GiB.
First, grow the RAID array. mdadm can grow the RAID array to a certain size, using the -G and -z options. The -z option can take a currently undocumented argument of max, which will resize the array to the maximum available space:
mdadm -G /dev/md0 -z max
`cat /proc/mdstat` and `mdadm -D /dev/md0` should show that the array is now using a 5 GiB device size.
Second, we need to enlarge the filesystem to match. Assuming that you have an ext3 filesystem on /dev/md0, and that you have mounted it, you can increase the size of the filesystem by using ext2online:
ext2online /dev/md0
After that command completes, you should see about 10 GiB of usable space.
Release Found: Red Hat Enterprise Linux 3 Update 7, Red Hat Enterprise Linux 4 Update 3
Symptom:
Certain HP hardware ships with an optional Broadcom 5715 NIC. The bnx2 module that ships with Red Hat Enterprise Linux 3 Update 7 and Red Hat Enterprise Linux 4 Update 3 contains a bug that makes restarting these NICs fail. The BCM5715 works during network installs and after the OS has intially loaded, but subsequent restarts of the NIC will fail.
The models currently effected include the BL460C G1 and BL480C G1 blades
Solution:
Red Hat Enterprise Linux 3 Update 8 (available July 18, 2006) and Red Hat Enterprise Linux 4 Update 4 (available August 7, 2006) will ship with fixes for this problem.
A temporary fix for those who have this problem before the release dates of the fixes should update their bnx2 modules from HP.
RHEL 3
http://h18023.www1.hp.com/support/files/server/us/download/24758.html
RHEL 4
http://h18023.www1.hp.com/support/files/server/us/download/24759.html
When the size of the data to be copied exceeds the size of physical memory, oom-killer starts randomly killing processes. This can be fixed by:
sysctl -w vm.lower_zone_protection 100
When lower_zone_protection is set to 100, it increases the free page threshold by 100 thereby starting page reclamation earlier and preventing NFS (Network File System) from getting far behind the kernel's memory demands.
For further details, refer to the Bugzilla entries:
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=193542
https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=163555
Release Found:
Red Hat Enterprise Linux
Listed below are few tips for identifying whether a cpu is dual-core:
Important points to note:
Examples of how to find the information on physical id and core id:
[root@harwich3i root]# grep phys /proc/cpuinfo physical id : 0 physical id : 8 physical id : 12 physical id : 4 physical id : 8 physical id : 12 physical id : 0 physical id : 4 physical id : 8 physical id : 12 physical id : 0 physical id : 4 physical id : 8 physical id : 12 physical id : 0 physical id : 4
[root@harwich3i root]# grep cores /proc/cpuinfo cpu cores : 2 (16 times)
[root@harwich3i root]# grep sib /proc/cpuinfo siblings : 4 (16 times)
[root@harwich3i root]# grep "core id" /proc/cpuinfo core id : 0 core id : 8 core id : 14 core id : 6 core id : 10 core id : 12 core id : 2 core id : 4 core id : 8 core id : 14 core id : 0 core id : 6 core id : 10 core id : 12 core id : 2 core id : 4
If the SMP kernel, ht or dual-core are not present the following fields will not be shown:
#ifdef CONFIG_SMP
if (smp_num_siblings > 1 || c->x86_num_cores > 1) {
int cpu = c - cpu_data;
seq_printf(m, "physical id\t: %d\n", phys_proc_id[c - cpu_data]);
seq_printf(m, "siblings\t: %d\n", cpus_weight(cpu_core_map[cpu]));
seq_printf(m, "core id\t\t: %d\n", cpu_core_id[c - cpu_data]);
seq_printf(m, "cpu cores\t: %d\n", c->x86_num_cores);
}
#endif
This means that it is either not a dual-core cpu, it is disabled, or the wrong kernel is being used.
After installing Red Hat Enterprise Linux 3 Update 8 with a single node configuration an a multi-node IBM x460 (x3950), and then merging into a two-node configuration; in order to use the storage adapter (Adaptec 9410 SAS or ServerRAID) located on the second node, the following steps must be taken:
#/sbin/new-kernel-pkg --install --mkinitrd `uname -r`
This will ensure that on the next boot, the device driver for the second node will be loaded automatically.
Manually load the module by running one of the following commands as root:
#/sbin/modprobe adp94xx
#/sbin/modprobe aacraid
The process of upgrading a Red Hat Network (RHN) Proxy Server is relatively simple. Note that the RHN Satellite Server must be upgraded to 4.x prior to an upgrade of the RHN Proxy Server to 4.x.
In the simplest terms an upgrade is as such:
up2date -uf
/sbin/service rhn-proxy stop
/sbin/service rhn-proxy start
While upgrading an RHN Proxy Server which is configured as a Monitoring Scout, be aware that any probes configured against this Scout will be lost during the upgrade process.This includes version 4.0.
# up2date -uf
# /sbin/service rhn-proxy stop
# /sbin/service rhn-proxy start
# /sbin/service httpd stop
# /sbin/service squid stop
# /sbin/service squid start
# /sbin/service httpd start
# /sbin/service rhn_auth_cache restart
# up2date -l
The goal of this section is to refactor the SSL build directory into a format compatible with RHN v3.7 or better products. A secondary goal is to create an SSL tar archive to be used during the installation of RHN Proxy 3.7 or better (RHN hosted connected installs only). In addition to this document, refer to the Client Configuration Guide for information on using and generating SSL certificates.
There are several scenarios:Refer to this article.
rhn-ssl-tool --gen-ca --dir ssl-build-temp <EXTRA OPTIONS>
rhn-ssl-tool --gen-server ^C #i.e. hit control-C at the password prompt
mv ssl-build/unknown ssl-build/cd ssl-build/ cat server.crt server.key > server.pem cd ../.. rhn-ssl-tool --gen-server --rpm-only
cd ssl-build/cat server.crt server.key > server.pem cd ../.. rhn-ssl-tool --gen-server --rpm-only
Refer Article 5410
Follow the recommendations of the Client Configuration Guide for storage of the ssl-build tree.
If steps 3, 4, 5, or 6 are followed there should now be a tar archive (new for version 3.7) in the /root/ssl-build/MACHINE-NAME/ directory in this format: rhn-org-httpd-ssl-archive-
Install RHN Proxy Server via it's top-level parent RHN service, which is either RHN Satellite or RHN hosted. Refer to the Proxy Installation Guide for instructions. Use the the hard-copy of /etc/rhn/rhn.conf for values to enter into the installer (especially traceback_mail, HTTP proxy settings, etc.). When asked, upload the ssl tar archive created previously (RHN hosted only).
Restore old RHN Proxy Server's custom package repository to the default location of an RHN Proxy Server 4.0:
restore /var/up2date/packages (or /var/spool/rhn-proxy if that applies) to /var/spool/rhn-proxy chmod 0750 /var/spool/rhn-proxy chown apache.apache /var/spool/rhn-proxy mkdir -m 0750 -p /var/spool/rhn-proxy/list chown apache.apache /var/spool/rhn-proxy/listUpdate, restart services, and test
up2date -uf /sbin/service rhn-proxy stop /sbin/service rhn-proxy start
Tunable parameters for heartbeat checks are in /proc/cluster/config/cman as hello_timer, deadnode_timer and max_retries. They are represented in seconds and take effect immediately. Any changes made should be carefully considered because only max_retries is safely changed on the fly whereas changes for hello_timer and deadnode_timer should be made after loading the cman module but before joining the cluster (cman_tool join), as they need to be the same across all nodes.
Note: If the value is set too low, the machine could stop working while doing something heavy processing
For more technical details:
http://www.redhat.com/archives/linux-cluster/2005-October/msg00070.html
The information provided in this article is for your information only. The origin of this information may be internal or external to Red Hat. While Red Hat attempts to verify the validity of this information before it is posted, Red Hat makes no express or implied claims to its validity.
This article is protected by the Open Publication License, V1.0 or later. Copyright © 2006 by Red Hat, Inc.