Oakies Blog Aggregator

connor_mc_d's picture

18.3 As easy as 1…2…3

Well, finally it’s here! 18c for on-premise installation so the world can all get stuck into the cool new features of the latest release on their own laptops Smile  At least that is what I’ll be doing!

Naturally as soon as I heard the news, I downloaded the software and got ready to set aside the day for installation and creation of an 18c database. But I didn’t need that long – I didn’t need that long at all. Just a few clicks and a few commands and there it was – my 18c database up and running.

Check out how easy it is with my three videos.

Software Installation

Listener Creation

Database Creation

It really is as easy as 1…2…3

Enjoy 18c !

martin.bach's picture

Little things worth knowing: Creating a RAC One Node database on the command line

This post is going to be super short, and mostly just a note to myself as I constantly forget how to create a RAC One database on the command line. This post is for 12.2.0.1 but should be similar on 12.1 (although I didn’t test!).

Provided you are licensed appropriately, this is probably the most basic way how you create an admin-managed RAC One database on Linux for use in a lab environment:

dbca -silent -createDatabase -gdbName RON -templateName gold_image01.dbc \
 -createAsContainerDatabase false -databaseConfigType RACONENODE \
 -RACOneNodeServiceName RON_SVC -totalMemory 1024 \
 -nodelist rac122node1,rac122node2 -storageType ASM \
 -datafileDestination '+DATA' -recoveryAreaDestination '+RECO' \
 -recoveryAreaSize 10240

This works for me, but most likely not for you :) And it’s certainly not suitable for a production deployment. Make sure to adapt the command as needed; I tend to create gold images for use with dbca, and this is one example.

The command itself should be fairly self-explanatory. If you are unsure about the meaning of the various options, have a look at the output of “dbca -help -createDatabase” and the official documentation/My Oracle Support. I learned the hard way that forgetting the “-nodelist” results in a single instance creation instead of an error message.

I didn’t find too many examples on the net, hope someone finds this useful.

Franck Pachot's picture

Release 18.0.0.0.0 Version 18.3.0.0.0 On-Premises binaries

By Franck Pachot

.
Good news, the latest Patchset for Oracle 12cR2 (which is not named patchset anymore, is actually called release 18c and numbered 18.0.0.0.0) is available for download on OTN. It is great because OTN download does not require access to Support and Software Updates. It is available to anybody under the Free Developer License Terms (basically development, testing, prototyping, and demonstrating for an application that is not in production and for non-commercial use). We all complained about the ‘Cloud First’ strategy because we were are eager to download the latest version. But the positive aspect of it is that we have now on OTN a release that has been stabilized after a few release updates. In the past, only the first version of the latest release was available there. Now we have one with many bug fixed.

Of course, I didn’t wait and I have tested 18c as soon as it was available on the Oracle Cloud thanks to the ACE Director program that provided me with some Cloud Credits. In this post, I’ll update my Cloud database to run it with the on-premises binary. Because that’s the big strength of Oracle: we can run the same software, 100% compatible, on the Cloud and on our own servers. There are some limitations in the features available, but technically it is the same software.

Oracle Cloud First

Here is my Cloud version of Oracle 18c installed on February (18.1) updated on April (18.2) and July (18.3):

SQLcl: Release 18.2 Production on Tue Jul 24 11:02:56 2018
 
Copyright (c) 1982, 2018, Oracle. All rights reserved.
 
Connected to:
Oracle Database 18c Enterprise Edition Release 18.0.0.0.0 - Production
Version 18.3.0.0.0
 
SQL> host $ORACLE_HOME/OPatch/opatch lspatches
28090523;Database Release Update : 18.3.0.0.180717 (28090523)
OPatch succeeded.
 
SQL> select banner from v$version;
 
BANNER
--------------------------------------------------------------------------------
Oracle Database 18c Enterprise Edition Release 18.0.0.0.0 - Production
 
SQL> select banner_full from v$version;
 
BANNER_FULL
----------------------------------------------------------------------------------------------------
Oracle Database 18c Enterprise Edition Release 18.0.0.0.0 - Production
Version 18.3.0.0.0
 
SQL> select banner_legacy from v$version;
 
BANNER_LEGACY
--------------------------------------------------------------------------------
Oracle Database 18c Enterprise Edition Release 18.0.0.0.0 - Production
 
SQL> exec dbms_qopatch.get_sqlpatch_status;
 
Patch Id : 27676517
Action : APPLY
Action Time : 18-APR-2018 20:44:50
Description : Database Release Update : 18.2.0.0.180417 (27676517)
Logfile : /u01/app/oracle/cfgtoollogs/sqlpatch/27676517/22097537/27676517_apply_CDB1_CDBROOT_2018Apr18_20_43_27.log
Status : SUCCESS
 
Patch Id : 28090523
Action : APPLY
Action Time : 18-JUL-2018 11:38:20
Description : Database Release Update : 18.3.0.0.180717 (28090523)
Logfile : /u01/app/oracle/cfgtoollogs/sqlpatch/28090523/22329768/28090523_apply_CDB1_CDBROOT_2018Jul18_11_36_38.log
Status : SUCCESS
 
PL/SQL procedure successfully completed.

LINUX.X64_180000_db_home.zip

I have installed the on-premises 18c available on OTN. The good things with the new releases are:

  • No need to extract installer files. Just unzip the Oracle Home and link the executable
  • This Oracle Home image already includes the latest Release Updates


SQLcl: Release 18.2 Production on Tue Jul 24 11:02:56 2018
 
Copyright (c) 1982, 2018, Oracle. All rights reserved.
 
Connected to:
Oracle Database 18c Enterprise Edition Release 18.0.0.0.0 - Production
Version 18.3.0.0.0
 
SQL> host $ORACLE_HOME/OPatch/opatch lspatches
27908644;UPDATE 18.3 DATABASE CLIENT JDK IN ORACLE HOME TO JDK8U171
27923415;OJVM RELEASE UPDATE: 18.3.0.0.180717 (27923415)
28090553;OCW RELEASE UPDATE 18.3.0.0.0 (28090553)
28090523;Database Release Update : 18.3.0.0.180717 (28090523)
OPatch succeeded.
 

We have 4 updates from July here for the following components:

  • The Database (28090523)
  • The Java in the Oracle Home, aka JDK (27908644)
  • The Java in the database, aka OJVM (27923415)
  • The clusterware component for the database to match the CRS, aka OCW (28090553)

So, now we have an image of the Oracle Home which already contains all the latest updates… except one:

$ cat $ORACLE_HOME/sqldeveloper/sqldeveloper/bin/version.properties
 
COMPANY=Oracle
PRODUCT=SQL Developer
VERSION=17.30003410937f
VER=17.3.2
VER_FULL=17.3.2.341.0937
BUILD_LABEL=341.0937
BUILD_NUM=341.0937
EDITION=

Unfortunately, that’s an old version of SQL Developer here, and with no SQLcl. Then just download this additional one and unzip it in the Oracle Home.

DataPatch

So, what happens when I open the database that I have created on 18.1 and patched with 18.2 and 18.3 RUs on the Oracle Cloud? There are two updates for the database (DBRU and OJVM). The DBRU is already there then DataPatch has only to apply the OJVM:

[oracle@VM183x dbhome_1]$ $ORACLE_HOME/OPatch/datapatch
SQL Patching tool version 18.0.0.0.0 Production on Tue Jul 24 10:57:55 2018
Copyright (c) 2012, 2018, Oracle. All rights reserved.
 
Log file for this invocation: /u01/app/oracle/cfgtoollogs/sqlpatch/sqlpatch_11104_2018_07_24_10_57_5 5/sqlpatch_invocation.log
 
Connecting to database...OK
Gathering database info...done
 
Note: Datapatch will only apply or rollback SQL fixes for PDBs
that are in an open state, no patches will be applied to closed PDBs.
Please refer to Note: Datapatch: Database 12c Post Patch SQL Automation
(Doc ID 1585822.1)
 
Bootstrapping registry and package to current versions...done
Determining current state...done
 
Current state of interim SQL patches:
Interim patch 27923415 (OJVM RELEASE UPDATE: 18.3.0.0.180717 (27923415)):
Binary registry: Installed
PDB CDB$ROOT: Not installed
PDB PDB$SEED: Not installed
PDB PDB1: Not installed
 
Current state of release update SQL patches:
Binary registry:
18.3.0.0.0 Release_Update 1806280943: Installed
PDB CDB$ROOT:
Applied 18.3.0.0.0 Release_Update 1806280943 successfully on 18-JUL-18 11.38.20.321353 AM
PDB PDB$SEED:
Applied 18.3.0.0.0 Release_Update 1806280943 successfully on 18-JUL-18 11.38.21.802495 AM
PDB PDB1:
Applied 18.3.0.0.0 Release_Update 1806280943 successfully on 18-JUL-18 11.38.23.230513 AM
 
Adding patches to installation queue and performing prereq checks...done
Installation queue:
For the following PDBs: CDB$ROOT PDB$SEED PDB1
No interim patches need to be rolled back
No release update patches need to be installed
The following interim patches will be applied:
27923415 (OJVM RELEASE UPDATE: 18.3.0.0.180717 (27923415))
 
Installing patches...
Patch installation complete. Total patches installed: 3
 
Validating logfiles...done
Patch 27923415 apply (pdb CDB$ROOT): SUCCESS
logfile: /u01/app/oracle/cfgtoollogs/sqlpatch/27923415/22239273/27923415_apply_CDB1_CDBROOT_2018Jul24_10_58_08. log (no errors)
Patch 27923415 apply (pdb PDB$SEED): SUCCESS
logfile: /u01/app/oracle/cfgtoollogs/sqlpatch/27923415/22239273/27923415_apply_CDB1_PDBSEED_2018Jul24_10_58_56. log (no errors)
Patch 27923415 apply (pdb PDB1): SUCCESS
logfile: /u01/app/oracle/cfgtoollogs/sqlpatch/27923415/22239273/27923415_apply_CDB1_PDB1_2018Jul24_10_58_56.log (no errors)
SQL Patching tool complete on Tue Jul 24 10:59:21 2018

Now here is the history of patches:

SQL> exec dbms_qopatch.get_sqlpatch_status;
 
Patch Id : 27676517
Action : APPLY
Action Time : 18-APR-2018 20:44:50
Description : Database Release Update : 18.2.0.0.180417 (27676517)
Logfile : /u01/app/oracle/cfgtoollogs/sqlpatch/27676517/22097537/27676517_apply_CDB1_CDBROOT_2018Apr18_20_43_27.log
Status : SUCCESS
 
Patch Id : 28090523
Action : APPLY
Action Time : 18-JUL-2018 11:38:20
Description : Database Release Update : 18.3.0.0.180717 (28090523)
Logfile : /u01/app/oracle/cfgtoollogs/sqlpatch/28090523/22329768/28090523_apply_CDB1_CDBROOT_2018Jul18_11_36_38.log
Status : SUCCESS
 
Patch Id : 27923415
Action : APPLY
Action Time : 24-JUL-2018 10:59:19
Description : OJVM RELEASE UPDATE: 18.3.0.0.180717 (27923415)
Logfile : /u01/app/oracle/cfgtoollogs/sqlpatch/27923415/22239273/27923415_apply_CDB1_CDBROOT_2018Jul24_10_58_08.log
Status : SUCCESS
 
PL/SQL procedure successfully completed.

This is all good. Despite the different release schedules, the level of software is exactly the same. And we can start on-premises on a release with low regression risk (18c like a patchset) but many fixes (several release updates). For the moment only the Linux port is there. The other platforms should come this summer.

 

Cet article Release 18.0.0.0.0 Version 18.3.0.0.0 On-Premises binaries est apparu en premier sur Blog dbi services.

pete.sharman's picture

That demned elusive archive log!

Introduction

With apologies to Emma Orczy again for stealing a line from “The Scarlet Pimpernel” … </p />
</p></div>
    <div class=»

pete.sharman's picture

That demned elusive archive log!

Introduction

With apologies to Emma Orczy again for stealing a line from “The Scarlet Pimpernel” … </p />
</p></div>
    <div class=»

dbakevlar's picture

Power BI 101- Logging and Tracing, Part II

So we went over locations and the basics of logging and tracing in Power BI.  I now want to know how to make more sense from the data.  In Oracle, we use a utility called TKProf, (along with others and a number of third party tools) to make sense of what comes from the logs.  SQL Server has Log Analytics and the profiler, but what can I do with Power BI?

First, let’s discuss what happens when we have actual activity.  In my first post, the system was pretty static.  This time I chose to open up a file with larger data refreshes from multiple sources, added tables, calculated columns and measures.  The one Access DB has over 10 million rows that is refreshed when I first open the PBIX file:

https://i0.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging.png?r... 300w, https://i0.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging.png?r... 768w, https://i0.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging.png?w... 1300w, https://i0.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging.png?w... 1950w" sizes="(max-width: 628px) 100vw, 628px" data-recalc-dims="1" />

Post loading, there’s a significant increase in number of MS Mashup Container, (calculations and measures) and msmdsrv, (data loading) logging:

https://i0.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging1.png?... 300w, https://i0.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging1.png?... 768w, https://i0.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging1.png?... 1300w" sizes="(max-width: 650px) 100vw, 650px" data-recalc-dims="1" />

Do I really want to go through all this data by hand?  BI is a reporting tool, so what if I bring them into Power BI?  Let’s start with the first MS Mashup Container log-

In Power BI, click on Get Data –> Text and change the file type to “All Files” in the explorer and go to the directory that contains the trace files:

C:\Users\\AppData\Local\Microsoft\Power BI Desktop\Traces\Performance

Remember that you will need to have “hidden items” set to be displayed to browse down to this folder.  Choose the files you wish to load in the directory and Power BI and choose a Customer delimiter of a quotes, (“) to separate the file.  This will load a file that will have a few columns you’ll need to remove that contain data like colons, nulls and other syntax from the file.  Once you’ve completed doing this, you most likely have a table with 15 columns of valuable data:

https://i1.wp.com/dbakevlar.com/wp-content/uploads/2018/07/columns_loggi... 145w" sizes="(max-width: 175px) 100vw, 175px" data-recalc-dims="1" />

I’ve renamed the columns to something more descriptive and I can now apply these changes and pull some value from the data.

Using the provided data, I can then produce a report that tells me about what types of processes are the largest users of resources and time.  I can provide reports to grant a visual on what’s going on in a Power BI environment.  The report is pretty straightforward-  Wait events against percentage of waits, Memory allocation over time, Time Waited and Wait Count.  These reports may seem really foreign for most data scientists, but for a DBA, it should resonate and provide them with ways they can offer assistance to the Power BI group in optimization.

https://i1.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging_bi.pn... 300w, https://i1.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging_bi.pn... 768w, https://i1.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging_bi.pn... 1300w, https://i1.wp.com/dbakevlar.com/wp-content/uploads/2018/07/logging_bi.pn... 1950w" sizes="(max-width: 650px) 100vw, 650px" data-recalc-dims="1" />

I can add hierarchy to this to drill down into interesting areas of waits and add more files, identifying each table by the file unique identifier and date that it came from going forward.  I expect my reports and my direction to look different from the direction many have taken with Power BI performance, but I wanted to demonstrate that optimization is always about time.  I admit fully that I’m still learning, but I also am approaching this from a database optimization perspective.  Please let me know your thoughts?

Happy hunting, folks!

 

 

 



Tags:  , ,


Del.icio.us



Facebook

TweetThis

Digg

StumbleUpon




Copyright © DBA Kevlar [Power BI 101- Logging and Tracing, Part II], All Right Reserved. 2018.

Richard Foote's picture

Announcement: Webinars for “Oracle Indexing Internals and Best Practices” Now Confirmed !!

Exciting News !! I can now confirm the dates for my first webinars of my fully updated and highly acclaimed “Oracle Indexing Internals and Best Practice” seminar. For details of all the extensive content covered in the webinars, please visit my Indexing Seminar page. The webinars will run for 4 hours each day, spanning a full week period […]

Franck Pachot's picture

Installing ZFS on OEL7 UEK4 for Docker storage

By Franck Pachot

.
The Oracle Database is fully supported on Docker according that Linux is Red Hat Enterprise Linux 7 or Oracle Enterprise Linux 7 with Unbreakable Enterprise 4. This is documented in MOS Note 2216342.1. Given the size of the Oracle database in GigaBytes even empty, the way it is installed at build with many file updates, and the per-block modifications of the datafiles, a block level copy-on-write filesystem is a must and deduplication and compression are appreciated. This makes ZFS a good option for the Docker storage driver, but also the external volumes. By the way, the Docker documentation about the storage drivers mention that zfs is a good choice for high-density workloads such as PaaS and this of course includes Database as a Service.

I’ve run this example on OEL 7.2 created in the the Oracle Cloud:
CaptureCreateInstance

kernel-uek-devel

We need to install the kernel headers. Of course, it is probably better to run a ‘yum update’ and reboot in order to run the latest kernel.
Here, I’m just installing the headers for the current kernel:

[root@localhost opc]# yum -y install kernel-uek-devel-$(uname -r)
...
Installed:
kernel-uek-devel.x86_64 0:4.1.12-112.14.13.el7uek
Dependency Installed:
cpp.x86_64 0:4.8.5-28.0.1.el7_5.1 gcc.x86_64 0:4.8.5-28.0.1.el7_5.1 glibc-devel.x86_64 0:2.17-222.el7
glibc-headers.x86_64 0:2.17-222.el7 kernel-headers.x86_64 0:3.10.0-862.9.1.el7 libdtrace-ctf.x86_64 0:0.8.0-1.el7
libmpc.x86_64 0:1.0.1-3.el7 mpfr.x86_64 0:3.1.1-4.el7
Dependency Updated:
glibc.x86_64 0:2.17-222.el7 glibc-common.x86_64 0:2.17-222.el7 libgcc.x86_64 0:4.8.5-28.0.1.el7_5.1
libgomp.x86_64 0:4.8.5-28.0.1.el7_5.1

DKMS

We need Dynamic Kernel Module Support to load ZFS modules. I had problems in the past with this so I install it step by step to verify that everything is ok. First, enable the EPEL repository:

[root@localhost opc]# yum install -y yum-utils
[root@localhost opc]# yum-config-manager --enable ol7_developer_EPEL

Then install DKMS:

[root@localhost opc]# yum -y install -y dkms
...
Installed:
dkms.noarch 0:2.4.0-1.20170926git959bd74.el7
Dependency Installed:
elfutils-default-yama-scope.noarch 0:0.170-4.el7 elfutils-libelf-devel.x86_64 0:0.170-4.el7
kernel-debug-devel.x86_64 0:3.10.0-862.9.1.el7 zlib-devel.x86_64 0:1.2.7-17.el7
Dependency Updated:
elfutils-libelf.x86_64 0:0.170-4.el7 elfutils-libs.x86_64 0:0.170-4.el7 zlib.x86_64 0:1.2.7-17.el7

Install ZFS repository

There is a zfs-release package that installs the /etc/yum.repos.d/zfs.repo:

[root@localhost opc]# sudo rpm -Uvh http://download.zfsonlinux.org/epel/zfs-release.el7_4.noarch.rpm
Retrieving http://download.zfsonlinux.org/epel/zfs-release.el7_4.noarch.rpm
warning: /var/tmp/rpm-tmp.yvRURo: Header V4 RSA/SHA256 Signature, key ID f14ab620: NOKEY
Preparing... ################################# [100%]
Updating / installing...
1:zfs-release-1-5.el7_4 ################################# [100%]

Basically, all it contains is the following enabled section:

[zfs]
name=ZFS on Linux for EL7 - dkms
baseurl=http://download.zfsonlinux.org/epel/7.4/$basearch/
enabled=1
metadata_expire=7d
gpgcheck=1
gpgkey=file:///etc/pki/rpm-gpg/RPM-GPG-KEY-zfsonlinux

Install ZFS

This is the important part, installing ZFS:

[root@localhost opc]# sudo yum install -y zfs
...
======================================================================================================================
Package Arch Version Repository Size
======================================================================================================================
Installing:
zfs x86_64 0.7.9-1.el7_4 zfs 413 k
Installing for dependencies:
kernel-devel x86_64 3.10.0-862.9.1.el7 ol7_latest 16 M
libnvpair1 x86_64 0.7.9-1.el7_4 zfs 30 k
libuutil1 x86_64 0.7.9-1.el7_4 zfs 35 k
libzfs2 x86_64 0.7.9-1.el7_4 zfs 130 k
libzpool2 x86_64 0.7.9-1.el7_4 zfs 591 k
lm_sensors-libs x86_64 3.4.0-4.20160601gitf9185e5.el7 ol7_latest 41 k
spl x86_64 0.7.9-1.el7_4 zfs 29 k
spl-dkms noarch 0.7.9-1.el7_4 zfs 456 k
sysstat x86_64 10.1.5-13.el7 ol7_latest 310 k
zfs-dkms noarch 0.7.9-1.el7_4 zfs 4.9 M

The most important is to check that the zfs module is installed correctly:

zfs.ko:
Running module version sanity check.
- Original module
- No original module exists within this kernel
- Installation
- Installing to /lib/modules/4.1.12-112.14.13.el7uek.x86_64/extra/

I’ve seen cases where it was not and then the module cannot load. You can also check:

[root@localhost opc]# dkms status
spl, 0.7.9, 4.1.12-112.14.13.el7uek.x86_64, x86_64: installed
zfs, 0.7.9, 4.1.12-112.14.13.el7uek.x86_64, x86_64: installed

If you have a problem (such as “modprobe: FATAL: Module zfs not found” when loading the module), check the status and maybe re-install it with:

dkms remove zfs/0.7.9 --all
dkms --force install zfs/0.7.9

If everything is ok, you can load the module:

[root@localhost opc]# /sbin/modprobe zfs
[root@localhost opc]#

Create a ZFS filesystem

If the ZFS module was not loaded you have this error:

[root@localhost opc]# zpool list
The ZFS modules are not loaded.
Try running '/sbin/modprobe zfs' as root to load them.

If it has been loaded correctly, you have no ZFS Storage Pool yet:

[root@localhost opc]# zpool list
no pools available

First I need to add a disk to my machine. Here I have only one disk created when I created the Compute Service:

[root@localhost opc]# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
xvdb 202:16 0 128G 0 disk
├─xvdb1 202:17 0 500M 0 part /boot
└─xvdb2 202:18 0 127.5G 0 part
├─vg_main-lv_root 249:0 0 123.5G 0 lvm /
└─vg_main-lv_swap 249:1 0 4G 0 lvm [SWAP]

I add a new disk in the Storage tab:
CaptureCreateStorage
And attach it and attach it to my Cloud Instance:
CaptureAttachStorage

Here is the new disk visible from the system:

[root@localhost opc]# lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
xvdb 202:16 0 128G 0 disk
├─xvdb1 202:17 0 500M 0 part /boot
└─xvdb2 202:18 0 127.5G 0 part
├─vg_main-lv_root 249:0 0 123.5G 0 lvm /
└─vg_main-lv_swap 249:1 0 4G 0 lvm [SWAP]
xvdc 202:32 0 120G 0 disk
&nbsp
[root@localhost opc]# ls -l /dev/xvdc /dev/block/202:32
lrwxrwxrwx 1 root root 7 Jul 19 15:05 /dev/block/202:32 -> ../xvdc
brw-rw---- 1 root disk 202, 32 Jul 19 15:05 /dev/xvdc

Here is where I add a ZFS Storage Pool for Docker:

[root@localhost opc]# zpool create -f zpool-docker -m /var/lib/docker /dev/xvdc
 
[root@localhost opc]# zpool status
pool: zpool-docker
state: ONLINE
scan: none requested
config:
 
NAME STATE READ WRITE CKSUM
zpool-docker ONLINE 0 0 0
xvdc ONLINE 0 0 0
 
[root@localhost opc]# zpool list
NAME SIZE ALLOC FREE EXPANDSZ FRAG CAP DEDUP HEALTH ALTROOT
zpool-docker 119G 118K 119G - 0% 0% 1.00x ONLINE -

And while I’m there I set some attributes to enable compression and deduplication. And as Docker writes to layers with 32k I/O I set the recordsize accordingly:

zfs set compression=on zpool-docker
zfs set dedup=on zpool-docker
zfs set recordsize=32k zpool-docker

Note that I attached if you reboot the instance you will have to attach the storage again and then run zpool import zpool-docker)

[root@localhost opc]# zpool import zpool-docker

Just to test that everything is ok, I install Docker as I did in a previous post:

[root@localhost opc]# yum-config-manager --add-repo https://download.docker.com/linux/centos/docker-ce.repo
[root@localhost opc]# yum-config-manager --enable ol7_addons
[root@localhost opc]# yum -y install docker-ce
[root@localhost opc]# systemctl start docker

Docker layers

I pull a small image and start a container on it:

[root@localhost opc]# docker run oraclelinux:7-slim

Here is the image and the ZFS dataset for its layer, mounted under /var/lib/docker/zfs:

[root@localhost opc]# docker image ls
REPOSITORY TAG IMAGE ID CREATED SIZE
oraclelinux 7-slim b1af4ba0cf19 12 days ago 117MB
 
[root@localhost opc]# docker inspect oraclelinux:7-slim | jq -r .[0].GraphDriver
 
{
"Data": {
"Dataset": "zpool-docker/fe31ff466872588506b1a3a3575c64d458beeb94d15bea593e5048237abf4fcc",
"Mountpoint": "/var/lib/docker/zfs/graph/fe31ff466872588506b1a3a3575c64d458beeb94d15bea593e5048237abf4fcc"
},
"Name": "zfs"
}

And here is the container layer:

[root@localhost opc]# docker container ls -a
 
CONTAINER ID IMAGE COMMAND CREATED STATUS PORTS NAMES
9eb7610c1fc5 oraclelinux:7-slim "/bin/bash" 6 minutes ago Exited (0) 6 minutes ago inspiring_shannon
 
[root@localhost opc]# docker inspect inspiring_shannon | jq -r .[0].GraphDriver
 
{
"Data": {
"Dataset": "zpool-docker/5d1022761b6dee28a25e21ec8c5c73d99d09863f11439bbf86e742856f844982",
"Mountpoint": "/var/lib/docker/zfs/graph/5d1022761b6dee28a25e21ec8c5c73d99d09863f11439bbf86e742856f844982"
},
"Name": "zfs"
}

If you don’t have jq just ‘yum install jq’. It is very convenient to filter and display the ‘inspect’ output.

We can see those datasets from ZFS list:

[root@localhost opc]# zfs list -o creation,space,snapshot_count,written -r | sort
 
CREATION NAME AVAIL USED USEDSNAP USEDDS USEDREFRESERV USEDCHILD SSCOUNT WRITTEN
Thu Jul 19 15:13 2018 zpool-docker 115G 126M 0B 964K 0B 125M none 964K
Thu Jul 19 15:38 2018 zpool-docker/fe31ff466872588506b1a3a3575c64d458beeb94d15bea593e5048237abf4fcc 115G 125M 0B 125M 0B 0B none 0
Thu Jul 19 15:39 2018 zpool-docker/5d1022761b6dee28a25e21ec8c5c73d99d09863f11439bbf86e742856f844982 115G 87K 0B 87K 0B 0B none 87K
Thu Jul 19 15:39 2018 zpool-docker/5d1022761b6dee28a25e21ec8c5c73d99d09863f11439bbf86e742856f844982-init 115G 46K 0B 46K 0B 0B none 0

Here, sorted by creation time, we see the datasets used by each layer. The initial files before having any image are less than 1MB. The image uses 125MB. The container creation has written 87KB and 46KB additional once running.

 

Cet article Installing ZFS on OEL7 UEK4 for Docker storage est apparu en premier sur Blog dbi services.

pete's picture

Oracle Security Training by Pete Finnigan in 2018

Are you worried about the data in your databases being stolen? GDPR has just become law across the EU and the UK and affects business in other countries that process EU citizens data. Maybe you store and process credit card....[Read More]

Posted by Pete On 19/07/18 At 02:04 PM

Franck Pachot's picture

Google Cloud Spanner – inserting data

By Franck Pachot

.
In a previous post I’ve created a Google Cloud Spanner database and inserted a few rows from the GUI. This is definitely not a solution fo many rows and here is a post about using the command line.

If I start the Google Shell from the icon on the Spanner page for my project, everything is set. But if I run it from elsewhere, using the https://console.cloud.google.com/cloudshell as I did in A free persistent Google Cloud service with Oracle XE I have to set the project:

franck_pachot@cloudshell:~$ gcloud config set project superb-avatar-210409
Updated property [core/project].
franck_pachot@superb-avatar-210409:~$

Instance

I create my Spanner instance with 3 nodes across the world:
¨
franck_pachot@superb-avatar-210409:~$ time gcloud spanner instances create franck --config nam-eur-asia1 --nodes=3 --description Franck
Creating instance...done.
 
real 0m3.940s
user 0m0.344s
sys 0m0.092s

Database

and Spanner database – created in 6 seconds:

franck_pachot@superb-avatar-210409:~$ time gcloud spanner databases create test --instance=franck
Creating database...done.
&nbssp;
real 0m6.832s
user 0m0.320s
sys 0m0.128s

Table

The DDL for table creation can also be run from there:

franck_pachot@superb-avatar-210409:~$ gcloud spanner databases ddl update test --instance=franck --ddl='create table DEMO1 ( ID1 int64, TEXT string(max) ) primary key (ID1)'
DDL updating...done.
'@type': type.googleapis.com/google.protobuf.Empty

I’m now ready to insert one million rows. Here is my table:

franck_pachot@superb-avatar-210409:~$ gcloud spanner databases ddl describe test --instance=franck
--- |-
CREATE TABLE DEMO1 (
ID1 INT64,
TEXT STRING(MAX),
) PRIMARY KEY(ID1)

Insert

The gcloud command line has a limited insert possibility:

franck_pachot@superb-avatar-210409:~$ time for i in $(seq 1 1000000) ; do gcloud beta spanner rows insert --table=DEMO1 --database=test --instance=franck --data=ID1=${i},TEXT=XXX${i} ; done
commitTimestamp: '2018-07-18T11:09:45.065684Z'
commitTimestamp: '2018-07-18T11:09:50.433133Z'
commitTimestamp: '2018-07-18T11:09:55.752857Z'
commitTimestamp: '2018-07-18T11:10:01.044531Z'
commitTimestamp: '2018-07-18T11:10:06.285764Z'
commitTimestamp: '2018-07-18T11:10:11.106936Z'
^C

Ok, let’s stop there. Calling a service for each row is not efficient with a latency of 5 seconds.

API

I’ll use the API from Python. Basically, a connection is a Spanner Client:

franck_pachot@superb-avatar-210409:~$ python3
Python 3.5.3 (default, Jan 19 2017, 14:11:04)
[GCC 6.3.0 20170118] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from google.cloud import spanner
>>> spanner_client = spanner.Client()
>>> instance = spanner_client.instance('franck')
>>> database = instance.database('test')
>>>

Batch Insert

With this I can send a batch of rows to insert. Here is the full Python script I used to insert one million, by batch of 1000 rows:

from google.cloud import spanner
spanner_client = spanner.Client()
instance = spanner_client.instance('franck')
database = instance.database('test')
for j in range(1000):
records=[]
for i in range(1000):
records.append((1+j*1000+i,u'XXX'+str(i)))
with database.batch() as batch:
batch.insert(table='DEMO1',columns=('ID1', 'TEXT',),values=records)

This takes 2 minutes:

franck_pachot@superb-avatar-210409:~$ time python3 test.py
 
real 2m52.707s
user 0m21.776s
sys 0m0.668s
franck_pachot@superb-avatar-210409:~$

If you remember my list of blogs on Variations on 1M rows insert that’s not so fast. But remember that rows are distributed across 3 nodes in 3 continents but here inserting with constantly increasing value have all batched rows going to the same node. The PRIMARY KEY in Google Spanner is not only there to declare a constraint but also determines the organization of data.

Query

The select can also be run from there from a read-only transaction called ‘Snapshot’ because it is doing MVCC consistent reads:

frank_pachot@superb-avatar-210409:~$ python3
Python 3.5.3 (default, Jan 19 2017, 14:11:04)
[GCC 6.3.0 20170118] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from google.cloud import spanner
>>> with spanner.Client().instance('franck').database('test').snapshot() as snapshot:
... results = snapshot.execute_sql('SELECT COUNT(*) FROM DEMO1')
... for row in results:
... print(row)
...
[1000000]

The advantage of the read-only transaction is that it can do consistent reads without locking. The queries executed in a read-write transaction have to acquire some locks in order to guarantee consistency when reading across multiple nodes.

Interleave

So, you can look at the PRIMARY KEY as a partition by range, and we have also reference partitioning with INTERLEAVE IN PARENT. This reminds me of the Oracle CLUSTER segment that is so rarely used because storing the tables separately is finally the better compromise on performance and flexibility for a multi-purpose database.

Here is my creation of DEMO2 where ID1 is a foreign key referencing DEMO1

franck_pachot@superb-avatar-210409:~$ time gcloud spanner databases ddl update test --instance=franck --ddl='create table DEMO2 ( ID1 int64, ID2 int64, TEXT string(max) ) primary key (ID1,ID2), interleave in parent DEMO1 on delete cascade'
DDL updating...done.
'@type': type.googleapis.com/google.protobuf.Empty
 
real 0m24.418s
user 0m0.356s
sys 0m0.088s

I’m now inserting 5 detail rows per each parent row:

from google.cloud import spanner
database = spanner.Client().instance('franck').database('test')
for j in range(1000):
records=[]
for i in range(1000):
for k in range(5):
records.append((1+j*1000+i,k,u'XXX'+str(i)+' '+str(k)))
with database.batch() as batch:
batch.insert(table='DEMO2',columns=('ID1','ID2','TEXT'),values=records)

This ran in 6 minutes.

Join (Cross Apply)

Here is the execution plan for

SELECT * FROM DEMO1 join DEMO2 using(ID1) where DEMO2.TEXT=DEMO1.TEXT

where I join the two tables and apply a filter on the join:
CaptureSpannerCrossApply

Thanks to the INTERLEAVE the join is running locally. Each row from DEMO1 (the Input of the Cross Apply) is joined with DEMO2 (the Map of Cross Apply) locally. Only the result is serialized. On this small number of rows we do not see the benefit from having the rows in multiple nodes. There are only 2 nodes with rows here (2 local executions) and probably one node contains most of the rows. The average time per node is 10.72 seconds and the elapsed time is 20.9 seconds, so I guess that one node ran un 20.9 seconds and the other in 1.35 only.

The same without the tables interleaved (here as DEMO3) is faster to insert but the join will be more complex where DEMO1 must be distributed to all nodes.
CaptureSpannerDistributedCrossApply
Without interleave, the input table of the local Cross Apply is a Batch Scan, which is actually like a temporary table distributed to all nodes (seems to have 51 chunks here), created by the ‘Create Batch’. This is called Distributed Cross Applied.

So what?

Google Spanner has only some aspects of SQL and Relational databases. But it is still, like the NoSQL databases, a database where the data model is focused at one use case only because the data model and the data organization have to be designed for specific data access.

 

Cet article Google Cloud Spanner – inserting data est apparu en premier sur Blog dbi services.

To prevent automated spam submissions leave this field empty.