Recent Posts by Alejandro Vargas

I've been adding storage cells to Exadatas for a long while, usually newer storage cells to a bit older Exadata. Yesterday I had the oposite case, moving six X3-2 cells to an X-7 Exadata. This mix of old and new storage cells is supported and can be done, but it took me a while to understand that the syntax of /opt/oracle.cellos/cell.conf had to remain as X-3. Usually I take the cell.conf from a working cell, and edit it to match the names and IP's on the cell to add, then run...

Synchronizing a MySQL slave database on the BDA is something I've done several times as part of the pre-requisites of a BDA upgrade. Is a simple task and there are two goods douments that provide a step by step guide to implement it. My favorite one is Doc 2152683.1, because almost does not affect online work, except for the time it takes to dump the Master database. How to Reset MySQL Slave Replication without Full BDA Cluster Downtime (Doc ID 2152683.1) If downtime is possible...

OEDACLI is a powerful administration tool for Exadata, one of the actions more appreciated by customers is adding a new database home, this is a summary of the steps required:     invoque oedacli (can be done on Windows or Linux)     on windows is located on the same directory from where you run the graphic version of OEDA     on Linux is located at /u01/install/linux-x64 on a physical installation     load the xml file          oedacli> load file name=test_Systems-x8m.xml        ...

Kerberos and Sentry help us defining access privileges for a limited set of object living on HDFS We can extend the control by using access control lists, ACL's  Requesting the ACL's on a directory will show us if "dfs.namenode.acls.enabled" is set to tru on HDFS: [root@linux-x64]# hdfs dfs -getfacl /user # file: /user # owner: hdfs # group: supergroup getfacl: The ACL operation has been rejected.  Support for ACLs has been disabled by setting dfs.namenode.acls.enabled to false. In...

Whenever we run a BDA upgrade we see as one of the prerequisites checks, to have the httpd service up and running on the first node of the cluster - The httpd service must be started on Node 1 of the cluster. It should be stopped on all other nodes. This is only needed during cluster upgrade or reconfiguration, but it is recommended to have it running at all times. The reason for that is because we will run yum from every node in the cluster pointing to the repository that is...

It may happen that when trying to reach the Exadata machine using the management network, the data center switch that connects to the Exadata Cisco switch get a 'connection rejected' error. On X7-2 port 48 of the Cisco switch is setup by default as ‘spanning tree network’ that set the BPDU* filter to disable, and communications between customer switch and the Cisco are rejected. To solve run this command on the Cisco: spanning tree BPDU filter en Enabling BPDU filtering in the...

Recently I had a BDA server that was refusing to come up. Every trial to run a start /SYS failed. We were able to identify this as a motherboard failure, then the motherboard was replaced and we were able to bring up the server using the ILOM. Back on track I've tried to setup the management network, but this failed because the eth0 NIC was missing. [root@bdanode01 ~]# ethtool -i eth0 Cannot get driver information: No such device But I've noticed we had eth1 available,...

Lately I had to help various customers to replace their old storage cells with the new X7-2 ones. There are huge benefits in doing this, the X7 has 25.5TB of flash, 12 x 10TB disks and 192 GB of DDR4 Memory.                       The question my customers asked the most was: Can we do the migration from our old storage to the new X7 without downtime and without risk? The answer was: YES! For doing this I've prepared and implemented a procedure that cover step by step how to migrate...

Setting passwordless ssh root connection using dcli is fast and simple and will easy later to execute commands on all servers using this utility. In order to do that you should have either: DNS resolution to all Database and Storage nodes OR have them registered in /etc/hosts 1) Create a parameter file that contains all the server names you want to reach via dcli, tipically we have a cell_group for storage cells, a dbs_group for database servers and an all_group for both of...