I’m working on Oracle database migration project where customer have chosen commodity x86 hardware with RHEL6 and EMC storage.

I’ve done many similar installations in the past and I always used the native MPIO in Linux (DM-Multipath) to load balance and failover I/O paths. This time however I’ve got EMC PowerPath doing the load balance and failover and got the native MPIO disabled. From my point of view it’s the same, whether I’ll be using /dev/emcpower* or /dev/mapper/* it’s the same. Obviously PowerPath has some advantages over the native MPIO which I really can’t tell yet. That’s a good paper from EMC giving a comparison between the native MPIO in different operating systems.

As mentioned before the aggregated logical names (pseudo names) with EMC PowerPath could be found under /dev/emcpowerX. I partitioned the disks with GPT tables and aligned the first partition to match the storage sector size. Also added to following line to udev rules to make sure my devices will get the proper permissions:

ACTION=="add", KERNEL=="emcpowerr1", OWNER:="oracle", GROUP:="dba", MODE="0600"

I restarted the server and then later udev to make sure ownership and permissions were picked up correctly. Upon running asmca to create ASM with the first disk group I got the following errors:

Configuring ASM failed with the following message:
One or more disk group(s) creation failed as below:
Disk Group DATA01 creation failed with the following message:
ORA-15018: diskgroup cannot be created
ORA-15031: disk specification '/dev/emcpowerr1' matches no disks
ORA-15025: could not open disk "/dev/emcpowerr1"
ORA-15056: additional error message

Well that’s strange, I’m sure the file had to correct permissions. However listing the file proved that it didn’t have the correct permissions. I repeated the process several times and always got the same result, you can use simple touch command to get the same result:

[root@testdb ~]# ls -al /dev/emcpowerr1
brw-rw---- 1 oracle dba 120, 241 Jan 23 12:35 /dev/emcpowerr1
[root@testdb ~]# touch /dev/emcpowerr1
[root@testdb ~]# ls -al /dev/emcpowerr1
brw-rw---- 1 root root 120, 241 Jan 23 12:35 /dev/emcpowerr1

Something was changing the ownership of the file and I didn’t know what. Well you’ll be no less surprised than I was to find that linux has a similar auditing framework as the Oracle database.

Auditctl will allow you to audit any file for any syscall run against it. In my case I would like to know which process is changing the ownership of my device file. Another helpful command is ausyscall whic allows you to map syscall names and numbers. In other words I would like to know what is the chmod syscall number on a 64bit platform (it does matter):

[root@testdb ~]# ausyscall x86_64 chmod --exact
90

Then I would like to set up auditing for all chmod calls against my device file:

[root@testdb ~]# auditctl -a exit,always -F path=/dev/emcpowerr1 -F arch=b64 -S chmod
[root@testdb ~]# touch /dev/emcpowerr1
[root@testdb ~]# tail -f /var/log/audit/audit.log
type=SYSCALL msg=audit(1422016631.416:4208): arch=c000003e syscall=90 success=yes exit=0 a0=7f3cfbd36960 a1=61b0 a2=7fff5c59b830 a3=0 items=1 ppid=60056 pid=63212 auid=4294967295 uid=0 gid=0 euid=0 suid=0 fsuid=0 egid=0 sgid=0 fsgid=0 tty=(none) ses=4294967295 comm="udevd" exe="/sbin/udevd" key=(null)
type=CWD msg=audit(1422016631.416:4208):  cwd="/"
type=PATH msg=audit(1422016631.416:4208): item=0 name="/dev/emcpowerr1" inode=28418 dev=00:05 mode=060660 ouid=54321 ogid=54322 rdev=78:f1 nametype=NORMAL
[root@testdb ~]# auditctl -D
No rules

Gotcha! So it was udev changing the permissions but why ?

I spent half day going through logs and tracing udev but couldn’t find anything.

At the end of the day I found an article by RHEL on which they had exactly the same problem. The solution was to have “add|change” into the ACTION directive instead of only “add”.

So here is the rule you need to have in order for UDEV to set a persistent ownership/permission on EMC PowerPath device files in RHEL 6:

[root@testdb ~]# cat /etc/udev/rules.d/99-oracle-asm.rules
ACTION=="add|change", KERNEL=="emcpowerr1", OWNER:="oracle", GROUP:="dba", MODE="0600"

Hope it helps and you don’t have to spent half day as I did.

Sve

It’s been a while since my last post but I was really busy working on a number of projects.

The purpose of this post is to highlight an issue I had while building a standby database. The environment we had – three 11.2.0.3 databases at host A (primary) and same were restored from backup on another host B (standby), both hosts were running Linux. It’s important to mention that both hosts were located in different Data Centers.

Once a standby database was mounted we would start shipping archive log files from the primary without adding it to the DataGuard Broker config as of that moment. We wanted to touch the production as little as possible and would add the database to the broker config just before doing the switchover. In the meanwhile we would manually recover the standby database to reduce the apply lag once the database is being added to the broker config. This approach worked fine for two of the databases but we got this error for the third one:

Fri Mar 13 13:33:43 2015
RFS[33]: Assigned to RFS process 29043
RFS[33]: Opened log for thread 1 sequence 29200 dbid -707326650 branch 806518278
CORRUPTION DETECTED: In redo blocks starting at block 20481count 2048 for thread 1 sequence 29200
Deleted Oracle managed file +RECO01/testdb/archivelog/2015_03_13/thread_1_seq_29200.8481.874244023
RFS[33]: Possible network disconnect with primary database
Fri Mar 13 13:42:45 2015
Errors in file /u01/app/oracle/diag/rdbms/testdb/testdb/trace/testdb_rfs_31033.trc:

Running through the trace file the first thing which I noticed was:

Corrupt redo block 5964 detected: BAD CHECKSUM

We already had two databases running from host A to host B so we rulled out the firewall issue. Then tried couple of other things – manually recovered the standby with incremental backup, recreated the standby, cleared all the redo/standby log groups but nothing helped. I found only one note in MOS with similar symptom for Streams in 10.2.

At the end the network admins were asked to check the config of the firewalls one more time. There were two firewalls – one where host A was located and another one where host B was located.

It turned out that the firewall at host A location had SQLnet class inspection enabled which was causing the corruption. The logs were successfully shipped from the primary database once this firewall feature was disabled. The strange thing was that we haven’t had any issues with the other two databases running on the same hosts, well what can I say – smart firewalls.

During a recent X5 installation I had to configure Link Aggregation Control Protocol (LACP) on the client network of the compute nodes. Although the ports were running at 10Gbits and default configuration of Active/Passive works perfectly fine the customer wanted even distribution of traffic and workload across their core switches.

Link Aggregation Control Protocol (LACP), also known as 802.3ad is a methods of combining multiple physical network connections into one logical connection to increase throughput and provide redundancy in case one of the links should fail. The protocol requires both – the server and the switch(es) to have the same settings to allow LACP to work properly.

To configure LACP on Exadata you need to change the bondeth0 parameters.

On each of the compute nodes open the following file:

/etc/sysconfig/network-scripts/ifcfg-bondeth0

and replace the line saying BONDING_OPTS with this one:

BONDING_OPTS="mode=802.3ad xmit_hash_policy=layer3+4 miimon=100 downdelay=200 updelay=5000 num_grat_arp=100"

and then restart the network interface:

ifdown bondeth0
ifup bondeth0
Determining if ip address 192.168.1.10 is already in use for device bondeth0...

You can check the status of the interface by query the proc filesystem. Make sure both interfaces are up and running at the same speed. The esential part to make sure the LACP is working is shown below:

cat /proc/net/bonding/bondeth0

802.3ad info
LACP rate: slow
Aggregator selection policy (ad_select): stable
Active Aggregator Info:
Aggregator ID: 1
Number of ports: 2
Actor Key: 33
Partner Key: 34627
Partner Mac Address: 00:23:04:ee:be:c8

cat /proc/net/bonding/bondeth0

802.3ad info
LACP rate: slow
Aggregator selection policy (ad_select): stable
bond bondeth0 has no active aggregator

With the release of Exadata X5 Oracle introduced elastic configurations and changed the process on how the initial configuration is performed. Back before you had to run applyconfig.sh which would go across the nodes and change all the settings according to your config. This script has now evolved and it’s called applyElasticConfig.sh which is part of OEDA (onecommand). During one of the recent deployments I ran into the below problem:

[root@node8 linux-x64]# ./applyElasticConfig.sh -cf Customer-exa01.xml

Applying Elastic Config...
Applying Elastic configuration...
Searching Subnet 172.16.2.x..........
5 live IPs in 172.16.2.x.............
Exadata node found 172.16.2.46.
Collecting diagnostics...
Errors occurred. Send /opt/oracle.SupportTools/onecommand/linux-x64/WorkDir/Diag-150512_160716.zip to Oracle to receive assistance.
Exception in thread "main" java.lang.NullPointerException
at oracle.onecommand.commandexec.utils.CommonUtils.getStackFromException(CommonUtils.java:1579)
at oracle.onecommand.deploy.cliXml.ApplyElasticConfig.doDaApply(ApplyElasticConfig.java:105)
at oracle.onecommand.deploy.cliXml.ApplyElasticConfig.main(ApplyElasticConfig.java:48)

Going through the logs we can see the following message:

2015-05-12 16:07:16,404 [FINE ][ main][ OcmdException:139] OcmdException from node node8.my.company.com return code = 2 output string: Unable to locate any IB switches... stack trace = java.lang.Throwable

The problem was caused because of IB switch names in my OEDA XML file were different to the one’s actually physically in the rack, actually the IB switch hostnames were missing from the hosts file. So if you ever run into this problem make sure your IB switch hosts file (/etc/hosts) has the correct hostname in the proper format:

#IP                 FQDN                      ALIAS
192.168.1.100       exa01ib01.local.net       exa01ib01

Also make sure to reboot the IB switch after any change of the hosts file.

I’m pleased to say that I will be speaking at the UKOUG Systems Event 2015, held at Cavendish Conference Center in London, 20 May 2015. My session “Oracle Exadata Meets Elastic Configurations” starts at 10:15 in Portland Suite. Here is the agenda of the UKOUG Systems Event.

In a month time I’ll be also speaking at the Spring Conference of the Bulgarian Oracle User Group. The conference will be held from 12^th to 14^th June, 2015 in hotel Novotel in Plovdiv, Bulgaria. I’ve got the conference opening slot at 11:00 in hall Moskva, my session topic is “Oracle Data Guard Fast-Start Failover: Live demo”. Here is the agenda of the conference.

I would like to thank EDBA for making this happen!

I was recently upgrading an Exadata 12.1.0.2 DBBP6 to DBBP7 and as usual I went for the latest opatch version which was 12.1.0.1.7 (Apr 2015) as of that time.

After running the opatchauto apply or opatchauto apply -analyze I got the following error:

System Configuration Collection failed: oracle.osysmodel.driver.sdk.productdriver.ProductDriverException: java.lang.NullPointerException
Exception in thread "main" java.lang.RuntimeException: java.io.IOException: Stream closed
        at oracle.opatchauto.gi.GILogger.writeWithoutTimeStamp(GILogger.java:450)
        at oracle.opatchauto.gi.GILogger.printStackTrace(GILogger.java:465)
        at oracle.opatchauto.gi.OPatchauto.main(OPatchauto.java:97)
Caused by: java.io.IOException: Stream closed
        at java.io.BufferedWriter.ensureOpen(BufferedWriter.java:98)
        at java.io.BufferedWriter.write(BufferedWriter.java:203)
        at java.io.Writer.write(Writer.java:140)
        at oracle.opatchauto.gi.GILogger.writeWithoutTimeStamp(GILogger.java:444)
        ... 2 more

opatchauto failed with error code 1.

This was a known problem caused by these bugs which are fixed in 12.1.0.1.8:
Bug 20892488 : OPATCHAUTO ANALYZE FAILING WITH GIPATCHINGHELPER::CREATESYSTEMINSTANCE FAILED
BUG 20857919 – LNX64-121023GIPSU:APPLY GIPSU FAILED WITH SYSTEM CONFIGURATION COLLECTION FAILED

As 12.1.0.1.8 is not yet available the workaround is to use lower version of opatch 12.1.0.1.6 which can be downloaded from this note:
Opatchauto Gives “System Configuration Collection Failed” Message (Doc ID 2001933.1)

You might run into the same problem if you are applying 12.1.0.2.3 PSU.

I’ve done several Exadata deployments recently and I have to say of all the components PDUs were hardest to configure. Important to notice that unlike earlier generations of Exadata the PDUs in X5 are Ehnanced PDUs and not Standard.

Reading the public documentation (Configuring the Power Distribution Units) it says that on PDUs with three power input leads you need to connect the middle power lead to the power source. Well I’ve done that many times and it didn’t NOT worked, the documentation says that the PDU should be accessible on 192.168.0.1. I believe the reason for that is because DHCP has been enabled by default and this can be easily confirmed by checking the LCD screen of the PDU. I even tried setting up DHCP server myself to make the PDU acquire IP but that didn’t worked either.

To configure the PDU you need to connect through serial management port. Nowadays there are no more laptops with serial ports so you will need USB to RS-232 DB9 Serial Adapter, I bought mine from Amazon. You will also need DB9 to RJ45 cable – these are quite popular and I’m sure you’ve seen before the blue Cisco console cable.

You need to connect the cable to SET MGT port of PDU and then establish terminal connection (you can use putty too) with the following settings:
9600 baud, 8 bit, 1 stop bit, no parity bit, no flow control

The username is admin and the password is adm1n.

Here are the commands you need to configure the PDU. Each network change requires reboot of the PDU:

Welcome to Oracle PDU

pducli->username: admin
pducli->password: *****
Login OK - Admin rights!
pducli->

set pdu_name=exa01pdu01
set systime_manual_date=2015-06-18
set systime_manual_time=12:45:00
set systime_ntp_server_enable=On
set systime_ntp_server=192.168.1.2
set systime_dst_enable=On
set net_ipv4_dhcp=Off
reset=yes

set net_ipv4_ipaddr=192.168.1.10
set net_ipv4_subnet=255.255.255.0
set net_ipv4_gateway=192.168.1.1
set net_ipv4_dns1=192.168.1.3
set net_ipv4_dns2=192.168.1.4
reset=yes

Regarding the network connectivity – the documentation says you need two additional cables from your management network. However if you have half or quarter rack you can plug-in the PDU network connections to the management/cisco switch. Make a note that if you ever plan to upgrade to full rack you will have to provide the two additional cables from your management network and disconnect PDUs from the management switch.

IMPORTANT: Make sure you don’t leave any active CLI sessions, otherwise you won’t be able to login remotely and will require data centre visit to reboot the PDU.

This is just a quick post to highlight a problem I had recently on another Exadata deployment.

For the most customers the management network on Exadata is routable and the DNS servers are accessible. However in a recent deployment for a financial organization this wasn’t the case and the storage servers were NOT able to reach the DNS servers. The customer provided a different set of DNS servers within the management network which were still able to resolve all the Exadata hostnames. If you encounter similar problem stop all cell services and run ipconf on each storage server to update the DNS servers.

On each storage server there is a service called cellwall (/etc/init.d/cellwall) which actually will run many checks and apply a lot of iptables rules. Here are couple of comments from the script to give you an idea:

# general lockdown from everything external, (then selectively permit)
  # general permissiveness (localhost: if you are in, you are in)
  # allow all udp traffic only from rdbms hosts on IPoIB only
      # allow DNS to work on all interfaces
      # open sport=53 only for DNS servers (mitigate remote-offlabel-port exploit)

and many more but you can check the script and see what it does OR run iptables -L -n to get all the iptables rules.

Here is some more information on how to change IP addresses on Exadata:
Changing IP addresses on Exadata Database Machine (Doc ID 1317159.1)

UPDATE: Thanks to Jason Arneil for pointing out that proper way to update the configuration of the cell.

I’ve done several Exadata deployments in the past two months and had to upgrade the Exadata storage software on half of them. Reason for that was because units shipped before May had their Exadata storage software version of 12.1.2.1.0.

The upgrade process of the database nodes ran fine but when I ran dbnodeupdate.sh -c for completing post upgrade steps I got an error that the system wasn’t on the expected Exadata release or kernel:

(*) 2015-06-01 14:21:21: Verifying GI and DB's are shutdown
(*) 2015-06-01 14:21:22: Verifying firmware updates/validations. Maximum wait time: 60 minutes.
(*) 2015-06-01 14:21:22: If the node reboots during this firmware update/validation, re-run './dbnodeupdate.sh -c' after the node restarts..
(*) 2015-06-01 14:21:23: Collecting console history for diag purposes

ERROR: System not on expected Exadata release or kernel, exiting


ERROR: Correct error, or to override run: ./dbnodeupdate.sh -c -q -t 12.1.2.1.1.150316.2

Indeed, the database node was running the new Exadata software but still using the old kernel (2.6.39-400.243) and dbnodeupdate was expecting me to run the new 2.6.39-400.248 kernel:

imageinfo:
Kernel version: 2.6.39-400.243.1.el6uek.x86_64 #1 SMP Wed Nov 26 09:15:35 PST 2014 x86_64
Image version: 12.1.2.1.1.150316.2
Image activated: 2015-06-01 12:27:57 +0100
Image status: success
System partition on device: /dev/mapper/VGExaDb-LVDbSys1

The reason for that was that the previous run of dbnodeupdate installed the new kernel package but failed to update grub.conf. The solution is to manually add the missing kernel entry to grub.conf and reboot the server to pick up the new kernel, here is a note for more information which by the time I had this problem was still internal:

Dbnodeupdate.sh Finishes With Error: System Not On Expected Exadata Release Or Kernel, Exiting (Doc ID 2007282.1)
Bug 20708183 – DOMU:GRUB.CONF KERNEL NOT ALWAYS UPDATED GOING TO 121211, NEW KERNEL NOT BOOTED

While deploying an X5 Full Rack recently it happened that the Grid Infrastructure Management Repository was not created by onecommand. The GIMR database was optional in 12.1.0.1 and became mandatory in 12.1.0.2 and should be automatically installed with Oracle Grid Infrastructure 12c release 1 (12.1.0.2). For unknown reason to me that didn’t happen and I had to create it manually. I’ve checked all the log files but couldn’t find any errors.  For reference the OEDA version used was Feb 2015 v15.050, image version on the Exadata was 12.1.2.1.0.141206.1.

To create the database login as the grid user and create file holding the following variables:

cat > /tmp/cfgrsp.properties
oracle.assistants.asm|S_ASMPASSWORD=[your ASM password]
oracle.assistants.asm|S_ASMMONITORPASSWORD=[your ASM password]

and run the following command:

GRID_HOME=/u01/app/12.1.0.2/grid
[oracle@exa01 ~]$ $GRID_HOME/cfgtoollogs/configToolAllCommands RESPONSE_FILE=/tmp/cfgrsp.properties

For reference, here is similar bug I found on MOS:
-MGMTDB Not Created When Using EM12c Provisioning (Doc ID 1983885.1)

This will be simple and short post on an issue I had recently. I got the following error while running the first step of onecommand – Validate Configuration File:

2015-07-01 12:31:03,712 [INFO  ][    main][     ValidationUtils:761] SUCCESS: NTP servers on machine exa01db02.local.net verified successfully
2015-07-01 12:31:03,713 [INFO  ][    main][     ValidationUtils:761] SUCCESS: NTP servers on machine exa01db01.local.net verified successfully
2015-07-01 12:31:03,714 [INFO  ][    main][     ValidationUtils:778] Following errors were found...
2015-07-01 12:31:03,714 [INFO  ][    main][     ValidationUtils:783] ERROR: Encountered error while running NTP validation error on host: exa01cel03.local.net
2015-07-01 12:31:03,714 [INFO  ][    main][     ValidationUtils:783] ERROR: Encountered error while running NTP validation error on host: exa01cel02.local.net
2015-07-01 12:31:03,714 [INFO  ][    main][     ValidationUtils:783] ERROR: Encountered error while running NTP validation error on host: exa01cel01.local.net

Right, so my NTP servers were accessible from the db nodes but not from the cells. When I queried the NTP server from the cells I got the following error:

# ntpdate -dv ntpserver1
1 Jul 09:00:09 ntpdate[22116]: ntpdate 4.2.6p5@1.2349-o Fri Feb 27 14:50:33 UTC 2015 (1)
Looking for host ntpserver1 and service ntp
host found : ntpserver1.local.net
transmit(172.16.1.100)
transmit(172.16.1.100)
transmit(172.16.1.100)
transmit(172.16.1.100)
transmit(172.16.1.100)
172.16.1.100: Server dropped: no data
server 172.16.1.100, port 123

Perhaps I should have mentioned that the cells have their own firewall (cellwall) which will only allow certain inbound/outbound traffic. During boot the script will build all the rules dynamically and apply them. Now the above error occurred because of two reasons:

A) The NTP servers were specified using hostname instead of IP addresses in OEDA
B) The management network was NOT available after the initial config (applyElasticConfig) was applied

Because of that cellwall was not able to resolve the NTP servers IP addresses and thus they were omitted from the firewall configuration. You can safely proceed with the deployment but if you want to get rid of the annoying message the solution is simply to restart the cell firewall – /etc/init.d/cellwall restart

This happened to me a month ago right after I applied DBBP7 on 12.1.0.2. For some reason the ora.crf resource didn’t start automatically:

CRS-5013: Agent "ORAROOTAGENT" failed to start process "/u01/app/12.1.0.2/grid/bin/osysmond" for action "start": details at "(:CLSN00008:)" in "/u01/app/oracle/diag/crs/exa01db01/crs/trace/ohasd_orarootagent_root.trc"
CRS-2674: Start of 'ora.crf' on 'exa01db01' failed

Checking the trace file for more details you can immediately spot where the problem is:

2015-06-04 10:35:51.156513 :CLSDYNAM:3286230784: [ ora.crf]{0:0:8275} [start] (:CLSN00008:)Utils:execCmd scls_process_spawn() failed 1
2015-06-04 10:35:51.156520 :CLSDYNAM:3286230784: [ ora.crf]{0:0:8275} [start] (:CLSN00008:) category: -1, operation: fail, loc: canexec2, OS error: 0, other: no exe permission, file [/u01/app/12.1.0.2/grid/bin/osysmond]

Indeed the osysmond is owned by the oracle user where it should be owned by root:

[root@exa01db01 ~]# ls -al /u01/app/12.1.0.2/grid/bin/osysmond
-rwxr-x--- 1 oracle oinstall 9441 Jun  4 10:42 /u01/app/12.1.0.2/grid/bin/osysmond

The fix for that is simple – you need to unlock and lock the GI:

[root@exa01db01 ~]# /u01/app/12.1.0.2/grid/crs/install/rootcrs.pl -unlock
[root@exa01db01 ~]# /u01/app/12.1.0.2/grid/crs/install/rootcrs.pl -patch

The osysmond has the correct permissions now and the resource ora.crf starts sucessfully:

[root@exa01db01 ~]# ls -al /u01/app/12.1.0.2/grid/bin/osysmond
-rwxr-x--- 1 root oinstall 9533 Jun  4 10:48 /u01/app/12.1.0.2/grid/bin/osysmond

For reference:
CRS-2672: Attempting to start ‘ora.crf’ on CRS-2674: Start of ‘ora.crf’ on failed CRS-2679: Attempting to clean ‘ora.crf’ on (Doc ID 1902447.1)

Exadata X5-2 and X4-8B racks are delivered with the “Enhanced” PDU metering units connected via the Cisco switch. Although the documentation says they should have static addresses, they don’t. You need to configure them manually using serial console connection, this is described in my earlier post here.

However if you forget to exit the serial console connection to the PDU and then try to login using SSH later you’ll get the following message:

login as: admin
admin@192.168.1.10's password:

CLI already in use!!!
Please try again later .....

Then someone has to go all the way to the data centre and reset the PDU or exit from the serial console.

That’s not really a problem but annoying issue I had with OEM 12c. Once a standby database is promoted, the database system for the same is showing as metric collections error OR Status Pending.

The standby database doesn’t need its own system since it will join the primary database system. The solution is to associate the standby database with the primary system and then remove the standby database system.

For example – we’ve got primary and standby databases – TESTDB_LON, TESTDB_RDG. Once promoted the following targets are also created in OEM – TESTDB_LON_sys and TESTDB_RDG_sys.

The second one will always be having status Pending:
Status Pending (Post Blackout)

The way to resolve that is to associate the standby database with the primary database system. I usually rename the primary database system as well to omit the location (LON and RDG):
– Go to the Targets -> Systems and choose the system you want to edit
– Then go to Database System -> Target Setup -> Edit system
– Rename the system name from TESTDB_LON_sys to TESTDB_sys
– Save changes
– Go to Database System again, Target Setup -> Edit system
– Click next to go to Step 2
– Add the standby database to the Standby Database Associations table
– Save changes

At this point we’ve got one system TESTDB_sys with two database members TESTDB_LON and TESTDB_RDG.

Next step is to remove the database system for the standby using emcli:

[oracle@oem12c ~]$ /opt/app/oracle/em12cr4/middleware/oms/bin/emcli login -username=sysman
Enter password :
Login successful

[oracle@oem12c ~]$ /opt/app/oracle/em12cr4/middleware/oms/bin/emcli delete_target -name="TESTDB_RDG_sys" -type="oracle_dbsys"
Target "TESTDB_RDG_sys:oracle_dbsys" deleted successfully

Now it’s all sorted and hopefully all targets are “green”.

The Oracle ASMFD (Filter Driver) was introduced in Oracle Database 12.1.0.2 and as of the moment it is available on Linux systems only.

Oracle ASM Filter Driver is a kernel module very much like the ASMLIB that resides in the I/O path of the Oracle ASM disks. It provides an interface between the Oracle binaries and the underlying operating environment.

Here are some of the features of ASMFD:

• Reject non-Oracle I/O

The ASM filter driver will reject write I/O operation issued by non-Oracle commands. This prevents non-Oracle applications from writing to ASM disks and protects ASM from accidental corruption.

• Device name persistence

Similarly to ASMLIB you don’t have to configure the device name persistence using UDEV.

• Faster node recovery

According to the documentation ASMFD allows Oracle Clusterware to perform node level fencing without a reboot. So in case of CSS is not running or nodes are fenced the Oracle stack will be restarted instead of node to be rebooted. This is greatly reduce the boot time as with some enterprise servers it might take up to 10 minutes to boot.

• Reduce OS resource usage

ASMFD exposes a portal device that can be used for all I/O on a particular host and thus decreasing the number of open file descriptors. With it each ASM process needs to have an open descriptor to each ASM disk. I’m not sure how much this will save you but might be useful in case you got hundreds of ASM disks.

• Thin Provisioning & Data Integrity

This is another new and cool feature which is very popular in the virtualization world. When enabled the disk space not in use can be returned to the array also known as thin-provisioning. This attribute can be set only if the ASM compatibility is greater than or equal to 12.1.0.0 and requires you to use ASMFD!

In a way ASMFD is a replacement of ASMLIB as it includes base-ASMLIB features. However ASMFD takes it one step further by protecting the ASM disks from non-oracle write I/O operations to prevent accidental damage. Unlike ASMLIB the ASMFD is installed with the Oracle Grid Infrastructure installation.

Brief history of ASM and the need of ASM Filter Driver

To understand ASMFD better we need to understand where the need comes from. It’s important to say that this is very specific to Linux as other platforms have other methods to fulfill the requirements. Because that’s not the purpose of this post and it’s too long I decide to keep it at the end of the post.

In Linux as in any other platform there is a user separation which implies access restrictions. In Linux we usually install Oracle Database under the oracle user and to do so we need to have writable access to the directories we plan to use. By default that would be /home/oracle/ and as you can imagine that’s not very handy, also you might want to install the database in separate partition or file system. For this reason the root user will create the required directories and change their ownership to oracle, that is usually /u01 or /opt.

That would work if you want to store your database files in a file system. However the traditional file systems were not designed for database files, they need to have a file system check on a regular basis and sometimes they might get corrupted. For that reason and performance perspective many people would move to RAW devices in the past. Another case would be if you want to run RAC – you’ll either need a cluster file system or RAW devices.

Historically with 9i and 10g we used to create RAW devices which are one to one mapping between a device file and a logical name. For example you would create partition on each device /dev/sda1, /deb/sdb1 and then map those to /dev/raw/raw1, /dev/raw/raw2 and so on. Additional because in Linux the device files are rebuild each time the system reboots you need to make sure the permissions and ownership are preserved and persist after system reboot. This was achieved by having additional rules in your last boot scripts (often rc.local). For other platforms like HP-UX for example one had to buy additional license (HP Service Guard extension for RAC) which would give you the ability to have a shared LVM groups across two or more servers.

However the support and maintenance of raw devices was really difficult and Oracle came up with the idea to create their own volume manager to simplify database administration and eliminate the need to manage thousands of database files – Automatic Storage Management, ASM for short. A simple description is that ASM is very sophisticated volume manager for Oracle data. ASM could also be used if you deploy RAC hence you don’t need cluster file systems or RAW devices anymore. Additionally it provides a redundancy so if you have JBOD you can use ASM to do the mirroring of the data. Another important feature is that you don’t need persistent device naming anymore. Upon start ASM will read all the disk drives specified by asm_diskstring and use the ones on which ASM header is found. Although ASM was released in 10.1 people were still using raw devices at the time because ASM was too new and unknown for many DBAs.

So ASM will logically group all the disks (LUNs presented from the storage) into what’s called ASM disk groups and because it’s using Oracle Managed Files you don’t really care anymore where your files are and what their names are. ASM is just another abstraction layer in the database file storage. ASM is available on all platforms so in a way it will standardize the administration of database files. Often the DBAs will also administer the ASM but it could be the storage team managing the ASM. You still had to make sure the device files have the correct permissions before ASM could use them, otherwise no diskgroup will be available hence database could not start.

At the same time back in 2004 Oracle released another product ASMLib which only purpose was to persist the device naming and preserve the device files permissions. I don’t want to go into details about ASMLib here but there is an old and very good post on ASMLib from Wim Coekaerts (HERE). Just to mention that ASMLib is also available under RHEL, more can be found HERE.

In the recent years many people like myself used UDEV to persist the permissions and ownership of the device files used by ASM. I really like to have one to one match between device files and ASM disk names for better understanding and ease any future troubleshooting.

ASM Filter Driver takes this one step further by introducing the features above. I can see people start using ASMFD to take advantage of the thin provisioning OR make sure no one will overwrite (by mistake) the ASM device files, yes this happens and it happened to me recently.

While working on another Exadata project recently I found that OEM12c agents on the compute nodes were installed on different locations on each of the three Exadatas. On one of them was under /home/oracle/agent, another one had them under /opt/oracle/agent and third one had them under /oracle/agent. Obviously this was not the standard and the agents had to be moved under /u01/app/oracle/agent. The only problem with that was that the three Exadatas were already discovered along with some database targets. Fortunately this wasn’t production yet but it would still require all the agents to be reinstalled and all targets rediscovered.

Fortunately there is an easier way to move the OEM management agents to new location without all the hassle of reinstalling agents and rediscovering agents. In the following example the agent was installed in /home/oracle/agent/ and I had to move it to /u01/app/oracle/agent/.

First you need to download the ConvertToStandalone.pl utility from 2021782.1 and then upload it to the server under /home/oracle

You need to create a list of plugins, otherwise the move process will fail:

[oracle@exa01db01 ~]$ /home/oracle/agent/core/12.1.0.5.0/perl/bin/perl /home/oracle/agent/core/12.1.0.5.0/sysman/install/create_plugin_list.pl -instancehome /home/oracle/agent/core/12.1.0.5.0

This will create a file /home/oracle/agent/plugins.txt which is used by the perl script later.

Export the following variables:

export OLD_AGENT_HOME=/home/oracle/agent/core/12.1.0.5.0
export ORACLE_HOME=/u01/app/oracle/agent/core/12.1.0.5.0

Another thing is you need to do is to modify the SBIN_MODIFIED_VERSION from 12.1.0.4.0. to 12.1.0.5.0 in /home/oracle/agent/agentimage.properties, otherwise the process will fail.

Then run the perl script which will migrate the agent home to the new location:

[oracle@exa01db01 ~]$ /home/oracle/agent/core/12.1.0.5.0/perl/bin/perl /home/oracle/ConvertToStandalone.pl -instanceHome /home/oracle/agent/agent_inst -newAgentBaseDir /u01/app/oracle/agent

Pay attention that the script accepts two arguments, instanceHome is the agent instance home directory e.g. /home/oracle/agent/agent_inst/ and the newAgentBaseDir is the new base dir for the agent /u01/app/oracle/agent/

After the command completes you need to run root.sh as root:

[oracle@exa01db01 ~]# /u01/app/oracle/agent/core/12.1.0.5.0/root.sh
Finished product-specific root actions.
/etc exists

Deinstall the old agent:

[oracle@exa01db01 ~]$ /home/oracle/agent/core/12.1.0.5.0/perl/bin/perl /home/oracle/agent/core/12.1.0.5.0/sysman/install/AgentDeinstall.pl -agentHome /home/oracle/agent/core/12.1.0.5.0

Finally remove the old agent directory where a log file from the deinstall process is left:

[oracle@exa01db01 ~]$ rm -rf /home/oracle/agent

The beauty of this process is that the script will create a blackout AGT_CNT_BLK_OUT on a node level and then stop the agent. It will then migrate the agent to the new home, start the agent and finally remove the blackout. The whole process takes less than five minutes.

Deployment of Exadata requires you to generate configuration using Oracle Exadata Deployment Assistant (OEDA). Within the same the default  ASM disk groups names are DBFS_DG, RECOC1 and DATAC1. I usually change those to RECO01 and DATA01 as others doesn’t make sense to me and the only place where I find the default ones is on Exadata.

I had an incident last year where the Exadata deployed half way through and names were left by default so I had to delete the configuration and start from scratch.

For my big surprise I got request recently where customer wanted to change RECO01 and DATA01 to RECOC1 and DATAC1! This was a pre-prod system, already deployed and having few databases running. The Exadata was X5-2 running ESS 12.1.2.1.2 and GI 12.1.0.2.

If this ever happens to you, here is what you need to do:

1. Rename grid disks.
2. Rename ASM disk groups and ASM disk names.
3. Modify all databases to point to the new disk groups.

Rename grid disks

Since grid disks names consists of the disk group name they need to be changed too. Although this is not mandatory I strongly recommend it to avoid any confusion in the future.

The grid disks can be renamed very easily using cellcli but they should NOT be in use by GI at that time. Thus Grid Infrastructure has to be stopped on all servers, stop GI as root:

[root@exa01db01 ~]# /u01/app/12.1.0.2/grid/bin/crsctl stop cluster -all

Then run the following magic command to get the list of all grid disks and replace the disk group names with the new ones:

[root@exa01db01 ~]# for i in `dcli -g cell_group -l root cellcli -e list griddisk | awk -F":" '{print $2'} | awk '{print $1}'`; do echo "cellcli -e alter griddisk $i name=$i"; done | grep -v DBFS |sed -e "s/RECO01/RECOC1/2" -e "s/DATA01/DATAC1/2"

You’ll get a long list of cellcli commands – 12 for each cell which you need to run on the cell locally.

Once it’s done start the GI again and make sure all disk groups are mounted successfully:

[root@exa01db01 ~]# /u01/app/12.1.0.2/grid/bin/crsctl start cluster

Rename ASM disk groups and ASM disk names

Next is to rename the disk groups. To do so they must be dismounted on ALL cluster nodes before running renamedg on a disk group. Connect to each ASM instance and dismount the disk groups:

SQL> alter diskgroup datac1 dismount;

Diskgroup altered.

SQL> alter diskgroup recoc1 dismount;

Diskgroup altered.

At this point you can run renamеdg to rename the disk groups, here is an example for the DATAC1 disk group:

[oracle@exa01db01 ~]$ renamedg -dgname DATA01 -newdgname DATAC1

Parsing parameters..
renamedg operation: -dgname DATA01 -newdgname DATAC1
Executing phase 1
Discovering the group
Checking for hearbeat...
Re-discovering the group
Generating configuration file..
Completed phase 1
Executing phase 2
Completed phase 2

Do the same for RECO01 and after that make sure that both disk groups can be mounted on all database nodes successfully, then dismount them again so you rename the ASM disk names. In general there is a command to rename all the disks (ALTER DISKGROUP XX RENAME DISKS ALL) but it will rename the disks to a name of the form diskgroupname_####, where #### is the disk number. However ASM disk names have different names on Exadata (RECO01_CD_01_EXA01CEL01) and that’s why we need to rename them manually.

To rename the disks the disk group has to be mounted in restricted mode (so only one node in the cluster can mount the disk group). Then run the below two statement to generate the new ASM disk names:

SQL> alter diskgroup datac1 mount restricted;

Diskgroup altered.

SQL> select 'alter diskgroup datac1 rename disk ''' || name || ''' to ''' || REPLACE(name,'DATA01','DATAC1') || ''';' from v$asm_disk where name like 'DATA%';

SQL> select 'alter diskgroup recoc1 rename disk ''' || name || ''' to ''' || REPLACE(name,'RECO01','RECOC1') || ''';' from v$asm_disk where name like 'RECO%';

Execute the alter statement generated by the above two statements and mount both disk groups on all database nodes again.

There is no command to add the disk group back to Oracle Restart. They will be automatically added first time they are mounted. However you need to remove the old disk group resources:

[oracle@exa01db01 ~]$ srvctl remove diskgroup -g DATA01
[oracle@exa01db01 ~]$ srvctl remove diskgroup -g RECO01

Modify all databases to point to the new disk groups

The last step is to change datafile/tempfile/redolog files on all databases to point to the new disk groups. Make sure you disable block change tracking and flashback as database might not open since the location of BCT has changed:

SQL> alter database disable block change tracking;
SQL> alter database flashback off;

Next create pfile from spfile and substitute all the occurences of RECO01 and DATA01 OR you can modify the spfile just before you shut the database. Let’s assume you have created pfile, make sure all the parameters refer to the new disk group names, here are the default ones that you need to check:

*.control_files
*.db_create_file_dest
*.db_create_online_log_dest_1
*.db_create_online_log_dest_2
*.db_recovery_file_dest

Start the database in mount state and generate all the alter statements for datafiles/tempfiles and redologs:

[oracle@exa01db01 ~]$ sqlplus -s / as sysdba
set heading off
set echo off
set pagesize 140
set linesize 140
spool /tmp/rename.sql

select 'alter database rename file ''' || name || ''' to ''' || REPLACE(name,'DATA01','DATAC1') || ''';' from v$datafile;
select 'alter database rename file ''' || name || ''' to ''' || REPLACE(name,'DATA01','DATAC1') || ''';' from v$tempfile;
select 'alter database rename file ''' || member || ''' to ''' || REPLACE(member,'DATA01','DATAC1')|| ''';' from v$logfile where member like '%DATA%';
select 'alter database rename file ''' || member || ''' to ''' || REPLACE(member,'RECO01','RECOC1')|| ''';' from v$logfile where member like '%RECO%';
exit

Start another sqlplus and run the spool file from the above operation (rename.sql). At this point you can open the database (alter database open;). Once the database is open make sure you enable block change tracking and flashback:

SQL> alter database enable block change tracking;
SQL> alter database flashback on;

Finally change the database dependencies and spfile location:

For 12c databases:

[oracle@exa01db01 dbs]$ srvctl modify database -d dbm01 -nodiskgroup
[oracle@exa01db01 dbs]$ srvctl modify database -d dbm01 -diskgroup "DATAC1,RECOC1"
[oracle@exa01db01 dbs]$ srvctl modify database -d dbm01 -spfile +DATAC1/DBM01/spfiledbm01.ora

For 11g databases:

[oracle@exa01db01 dbs]$ srvctl modify database -d dbm01 -z
[oracle@exa01db01 dbs]$ srvctl modify database -d dbm01 -x "DATAC1,RECOC1"
[oracle@exa01db01 dbs]$ srvctl modify database -d dbm01 -p +DATAC1/DBM01/spfiledbm01.ora

It’s this time of the year again when one of the biggest and last for the year Oracle User Groups is being held and that is UK Oracle User Group Conference.

I’m very grateful and proud that I’ll be speaking on this great conference, here are my talks:

Presentation Title: Oracle Exadata Meets Elastic Configurations
Description: With the release of Exadata X5 Oracle announced Elastic configuration to allow mixed number of db and cell nodes. This session will go through the implementation process of a X5 having two db nodes and four cells.
Date: Monday 7th December
Time: 14:10 – 15:00
Hall: Media Suite B

Presentation Title: Oracle DataGuard Fast-Start Failover: Live Demo
Description: Come and see a live demo of Oracle Fast-start failover and why a private bank moved from RAC to FSFO.
Date: Tuesday 8th December
Time: 16:30 – 17:20
Hall: Hall 11A

See you there!

This is a quick post to highlight two issues I had with Oracle dNFS. Both relate to wrong entries in the oranfstab file.

One might encounter ORA-00600 during database creation:

DBCA_PROGRESS : 7%
DBCA_PROGRESS : 8%
ORA-01501: CREATE DATABASE failed
ORA-00600: internal error code, arguments: [KSSRMP1], [], [], [], [], [], [], [], [], [], [], []
ORA-17502: ksfdcre:3 Failed to create file /oracle/ORCL/database/ORCL/controlfile/o1_mf_%u_.ctl

This was caused by having a wrong entry in oranfstab – there was difference between fstab and oranfstab for the same record:

server: zfs01
path: 192.168.10.100
export: /export/OTHERDB/database mount: /oracle/ORCL/database

The second issue was that the database wasn’t using dNFS. Simple query of v$dnfs_servers will return no rows and there were no errors in the alertlog. However looking around the tracefiles one can easily spot the following repetitive error in all trace files:

KGNFS_NFSPROC3_MNT FAIL 13

This was caused by trying to mount a share we don’t have access to or non existing share from the NFS server:

server: zfs01
path: 192.168.10.100
export: /export/NON_EXIST/database mount: /oracle/ORCL/database

The issue was fixed after correcting the typos in the oranfstab file and resting the database.

The bottom line is make sure that fstab and oranfstab match and have correct entries.

I was installing another Exadata last month when I got an error on create cell disks step. I’ve seen the same error before when I was extending two to three rack Exadata configuration but thought it was one-off.

The cell disk creation failed as below:

[root@exa01db01 linux-x64]# ./install.sh -cf Customer-exa01.xml -s 8

 Initializing
 Executing Create Cell Disks
 Checking physical disks for errors before creating celldisks.........................
 Restarting cell services....................................................
 ERROR:

 Stopping the RS, CELLSRV, and MS services...
 The SHUTDOWN of services was successful.
 Starting the RS, CELLSRV, and MS services...
 Getting the state of RS services...  running
 Starting CELLSRV services...
 The STARTUP of CELLSRV services was not successful.
 CELL-01533: Unable to validate the IP addresses from the cellinit.ora file because the IP addresses may be down or misconfigured.
 Starting MS services...
 The STARTUP of MS services was successful.
 ERROR:

Going through the cell configuration is obvious why the process failed. The cell still had the default name and the IP addresses that the cell services should use are still the default ones:

CellCLI> list cell detail
         name:                   ru02
         ipaddress1:             192.168.10.1/24
         ipaddress2:             192.168.10.2/24
         cellsrvStatus:          stopped
         msStatus:               running
         rsStatus:               running

In short when you see an error like the one below then your ipaddress1 and/or ipaddress2 fields are most probably wrong:

         2       2015-12-15T17:57:03+00:00       critical        "ORA-00700: soft internal error, arguments: [main_6a], [3], [IP addresses in cellinit.ora not operational], [], [], [], [], [], [], [], [], []"

The solution to that is simple. You need to alter the cell name and IP addresses manually:

CellCLI> alter cell name=exa01cel02a,ipaddress1='192.168.10.13/22',ipaddress2='192.168.10.14/22'
Network configuration altered. Please issue the following commands as root to restart the network and open IB stack:
service openibd restart
service network restart
A restart of all services is required to put new network configuration into effect. MS-CELLSRV communication may be hampered until restart.
Cell exa01cel02a successfully altered

CellCLI> alter cell restart services all

Make sure all cells are fixed and re-run the onecommand step, this time it will succeed:

 Successfully completed execution of step Create Cell Disks [elapsed Time [Elapsed = 128338 mS [2.0 minutes] Thu Dec 17 14:26:59 GMT 2015]]

I’ve checked some older deployments and it’s the same step which should change the cell name and restart the cell services. For some reason this didn’t happened for me. For both deployments I used OEDA v15.300 (Oct 2015) so this might be a bug in this version.