Tuesday Jan 14, 2014

Configuration Applications / OS's for 1MB ZFS Block Sizes

The latest release of the ZFS Storage Appliance, 2013.1.1.1, introduces 1MB block sizes for shares. This is a deferred update that can only be enabled inside of Maintenance → System. You can edit individual Filesystems or LUNs from within 'Shares' to enable the 1MB support (database record size).

This new feature may need additional tweaking on all connected servers to fully realize significant performance gains. Most operating systems currently do not support a 1MB transfer size by default. This can be very easily spotted within analytics by breaking down your expected protocol by IO size. As an example, let's look at a fibre channel workload being generated by an Oracle Linux 6.5 server:

Example


The IO size is sitting at 501K, a very strange number that's eerily close to 512K. Why is this a problem? Well, take a look at our backend disks:

Our disk IO size (block size) is heavily fragmented! This causes our overall throughput to nosedive.

2GB/s is okay, but we can do better if our buffer size was 1MB on the host side.

Fixing the problem


Fibre Channel

Solaris
# echo 'set maxphys=1048576' > /etc/system

Oracle Linux 6.5 uek3 kernel (previous releases do not support 1MB sizes for multipath)
# echo 1024 > /sys/block/dm*/queue/max_sectors_kb 

or create a permanent udev rule:
# vi /etc/udev/rules.d/99-zfssa.rules
ACTION=="add", SYSFS{vendor}=="SUN", SYSFS{model}=="*ZFS*", ENV{ID_FS_USAGE}!="filesystem", ENV{ID_PATH}=="*-fc-*", RUN+="/bin/sh -c 'echo 1024 > /sys$DEVPATH/queue/max_sectors_kb'"

Windows

QLogic [qlfc]
C:\> qlfcx64.exe -tsize /fc /set 1024

Emulex [HBAnyware]
set ExtTransferSize = 1

Please see MOS Note 1640013.1 for configuration for iSCSI and NFS.

Results


After re-running the same FC workload with the correctly set 1MB transfer size, I can see the IO size is now where it should be.

This has a drastic impact on the block sizes being allocated on the backend disks:

And an even more drastic impact on the overall throughput:

A very small tweak resulted in a 5X performance gain (2.1GB/s to 10.9GB/s)! Until 1MB is the default for all physical I/O requests, expect to make some configuration changes on your underlying OS's.

System Configuration


Storage

  • 1 x Oracle ZS3-4 Controller
  • 2013.1.1.1 firmware
  • 1TB DRAM
  • 4 x 16G Fibre Channel HBAs
  • 4 x SAS2 HBAs
  • 4 x Disk Trays (24 4TB 7200RPM disks each)
Servers
  • 4 x Oracle x4170 M2 servers
  • Oracle Linux 6.5 (3.8.x kernel)
  • 16G DRAM
  • 1 x 16G Fibre Channel HBA

Workload


Each Oracle Linux server ran the following vdbench profile running against 4 LUNs:

sd=sd1,lun=/dev/mapper/mpatha,size=1g,openflags=o_direct,threads=128
sd=sd2,lun=/dev/mapper/mpathb,size=1g,openflags=o_direct,threads=128
sd=sd3,lun=/dev/mapper/mpathc,size=1g,openflags=o_direct,threads=128
sd=sd4,lun=/dev/mapper/mpathd,size=1g,openflags=o_direct,threads=128

wd=wd1,sd=sd*,xfersize=1m,readpct=70,seekpct=0
rd=run1,wd=wd1,iorate=max,elapsed=999h,interval=1
This is a 70% read / 30% write sequential workload.

Tuesday Sep 17, 2013

ZFS Storage at Oracle OpenWorld 2013

Join my colleagues and myself at this year's Oracle OpenWorld. I have a session, hands-on-lab, and demo being held in and around Moscone. These are all heavily focused on 12c and ZFS analytics.

HOL10103 - Managing ZFS Storage Inside Oracle Database 12c Environments 

September 23, (Monday) 10:45 AM - Marriott Marquis - Salon 10A
CON2846 - Oracle Use and Best Practices for High-Performance Cloud Storage 

September 23, (Monday) 12:15 PM - Westin San Francisco - Franciscan II
DEMO3619 - Maintaining the Performance of Your Cloud Infrastructure

Moscone South Lower Level, SC-152

Wednesday Jul 10, 2013

Solaris 11 IPoIB + IPMP

I recently needed to create a two port active:standby IPMP group to be served over Infiniband on Solaris 11. Wow that's a mouthful of terminology! Here's how I did it:

List available IB links

[root@adrenaline ~]# dladm show-ib
LINK         HCAGUID         PORTGUID        PORT STATE  PKEYS
net5         21280001CF4C96  21280001CF4C97  1    up     FFFF
net6         21280001CF4C96  21280001CF4C98  2    up     FFFF
Partition the IB links. My pkey will be 8001.
[root@adrenaline ~]# dladm create-part -l net5 -P 0x8001 p8001.net5
[root@adrenaline ~]# dladm create-part -l net6 -P 0x8001 p8001.net6
[root@adrenaline ~]# dladm show-part
LINK         PKEY  OVER         STATE    FLAGS
p8001.net5   8001  net5         unknown  ----
p8001.net6   8001  net6         unknown  ----
Create test addresses for the newly created datalinks
[root@adrenaline ~]# ipadm create-ip p8001.net5
[root@adrenaline ~]# ipadm create-addr -T static -a 192.168.1.101 p8001.net5/ipv4
[root@adrenaline ~]# ipadm create-ip p8001.net6
[root@adrenaline ~]# ipadm create-addr -T static -a 192.168.1.102 p8001.net6/ipv4
[root@adrenaline ~]# ipadm show-addr
ADDROBJ           TYPE     STATE        ADDR
p8001.net5/ipv4   static   ok           192.168.1.101/24
p8001.net6/ipv4   static   ok           192.168.1.102/24
Create an IPMP group and add the IB datalinks
[root@adrenaline ~]# ipadm create-ipmp ipmp0
[root@adrenaline ~]# ipadm add-ipmp -i p8001.net5 -i p8001.net6 ipmp0
Set one IB datalink to standby
[root@adrenaline ~]# ipadm set-ifprop -p standby=on -m ip p8001.net6
Assign an IP address to the IPMP group
[root@adrenaline ~]# ipadm create-addr -T static -a 192.168.1.100/24 ipmp0/v4
That's it! Final checks:
[root@adrenaline ~]# ipadm
NAME              CLASS/TYPE STATE        UNDER      ADDR
ipmp0             ipmp       ok           --         --
   ipmp0/v4       static     ok           --         192.168.1.100/24
p8001.net5        ip         ok           ipmp0      --
   p8001.net5/ipv4 static    ok           --         192.168.1.101/24
p8001.net6        ip         ok           ipmp0      --
   p8001.net6/ipv4 static    ok           --         192.168.1.102/24

[root@adrenaline ~]# ping 192.168.1.100
192.168.1.100 is alive

Monday Mar 04, 2013

Configuring a Basic DNS Server + Client in Solaris 11

Configuring the Server
The default install of Solaris 11 does not come with a DNS server, but this can be added easily through IPS like so:
[paulie@griff ~]$ sudo pkg install service/network/dns/bind
Before enabling this service, the named.conf file needs to be modified to support the DNS structure. Here's what mine looks like:
[paulie@griff ~]$ cat /etc/named.conf
options {
        directory       "/etc/namedb/working";
        pid-file        "/var/run/named/pid";
        dump-file       "/var/dump/named_dump.db";
        statistics-file "/var/stats/named.stats";
        forwarders { 208.67.222.222; 208.67.220.220; };
};

zone "hillvalley" {
        type master;
        file "/etc/namedb/master/hillvalley.db";
};

zone "1.168.192.in-addr.arpa" {
        type master;
        file "/etc/namedb/master/1.168.192.db";
};
My forwarders use the OpenDNS servers, so any request that the local DNS server can't process goes through there. I've also setup two zones: hillvalley.db for my forward zone and 1.168.192.db for my reverse zone. We need both for a proper configuration. We also need to create some directories to support this file:
[paulie@griff ~]$ sudo mkdir /var/dump
[paulie@griff ~]$ sudo mkdir /var/stats
[paulie@griff ~]$ sudo mkdir -p /var/run/namedb
[paulie@griff ~]$ sudo mkdir -p /etc/namedb/master
[paulie@griff ~]$ sudo mkdir -p /etc/namedb/working
Now, let's populate the DNS server with a forward and reverse file.

Forward file
[paulie@griff ~]$ cat /etc/namedb/master/hillvalley.db 
$TTL 3h
@       IN      SOA     griff.hillvalley. paulie.griff.hillvalley. (
        2013022744 ;serial (change after every update)
        3600 ;refresh (1 hour)
        3600 ;retry (1 hour)
        604800 ;expire (1 week)
        38400 ;minimum (1 day)
)

hillvalley.     IN      NS      griff.hillvalley.

delorean        IN      A       192.168.1.1   ; Router
biff            IN      A       192.168.1.101 ; NFS Server
griff           IN      A       192.168.1.102 ; DNS Server
buford          IN      A       192.168.1.103 ; LDAP Server
marty           IN      A       192.168.1.104 ; Workstation
doc             IN      A       192.168.1.105 ; Laptop
jennifer        IN      A       192.168.1.106 ; Boxee
lorraine        IN      A       192.168.1.107 ; Boxee
Reverse File
[paulie@griff ~]$ cat /etc/namedb/master/1.168.192.db 
$TTL 3h
@       IN      SOA     griff.hillvalley. paulie.griff.hillvalley. (
        2013022744 ;serial (change after every update)
        3600 ;refresh (1 hour)
        3600 ;retry (1 hour)
        604800 ;expire (1 week)
        38400 ;minimum (1 day)
)

        IN      NS      griff.hillvalley.

1       IN      PTR     delorean.hillvalley.    ; Router
101     IN      PTR     biff.hillvalley.        ; NFS Server
102     IN      PTR     griff.hillvalley.       ; DNS Server
103     IN      PTR     buford.hillvalley.      ; LDAP Server
104     IN      PTR     marty.hillvalley.       ; Workstation
105     IN      PTR     doc.hillvalley.         ; Laptop
106     IN      PTR     jennifer.hillvalley.    ; Boxee
107     IN      PTR     lorraine.hillvalley.    ; Boxee
For referencing how these files works:
  • paulie is the admin user account name
  • griff is the hostname of the DNS server
  • hillvalley is the domain name of the network
  • I love BTTF
Feel free to tweak this example to match your own network. Finally, enable the DNS service and check that it's online:
[paulie@griff ~]$ sudo svcadm enable dns/server
[paulie@griff ~]$ sudo svcs | grep dns/server
online         22:32:20 svc:/network/dns/server:default
Configuring the Client
We will need the IP address (192.168.1.102), hostname (griff), and domain name (hillvalley) to configure DNS with these commands:
[paulie@buford ~]$ sudo svccfg -s network/dns/client setprop config/nameserver = net_address: 192.168.1.102
[paulie@buford ~]$ sudo svccfg -s network/dns/client setprop config/domain = astring: hillvalley
[paulie@buford ~]$ sudo svccfg -s network/dns/client setprop config/search = astring: hillvalley
[paulie@buford ~]$ sudo svccfg -s name-service/switch setprop config/ipnodes = astring: '"files dns"'
[paulie@buford ~]$ sudo svccfg -s name-service/switch setprop config/host = astring: '"files dns"'
Verify the configuration is correct:
[paulie@buford ~]$ svccfg -s network/dns/client listprop config
config                      application        
config/value_authorization astring     solaris.smf.value.name-service.dns.client
config/nameserver          net_address 192.168.1.102
config/domain              astring     hillvalley
config/search              astring     hillvalley
And enable:
[paulie@buford ~]$ sudo svcadm enable dns/client
Now we need to test that the DNS server is working using both forward and reverse DNS lookups:
[paulie@buford ~]$ nslookup lorraine
Server:         192.168.1.102
Address:        192.168.1.102#53

Name:   lorraine.hillvalley
Address: 192.168.1.107

[paulie@buford ~]$ nslookup 192.168.1.1
Server:         192.168.1.102
Address:        192.168.1.102#53

1.1.168.192.in-addr.arpa        name = delorean.hillvalley.

Thursday Feb 21, 2013

Configuring a Basic LDAP Server + Client in Solaris 11

Configuring the Server
Solaris 11 ships with OpenLDAP to use as an LDAP server. To configure, you're going to need a simple slapd.conf file and an LDIF schema file to populate the database. First, let's look at the slapd.conf configuration:
# cat /etc/openldap/slapd.conf
include         /etc/openldap/schema/core.schema
include         /etc/openldap/schema/cosine.schema
include         /etc/openldap/schema/inetorgperson.schema
include         /etc/openldap/schema/nis.schema

pidfile         /var/openldap/run/slapd.pid
argsfile        /var/openldap/run/slapd.args

database        bdb
suffix          "dc=buford,dc=hillvalley"
rootdn          "cn=admin,dc=buford,dc=hillvalley"
rootpw          secret
directory       /var/openldap/openldap-data
index           objectClass     eq
You may want to change the lines suffix and rootdn to better represent your network naming schema. My LDAP server's hostname is buford and domain name is hillvalley. You will need to add additional domain components (dc=) if the name is longer. This schema assumes the LDAP manager will be called admin. Its password is 'secret'. This is in clear-text just as an example, but you can generate a new one using slappasswd:
[paulie@buford ~]$ slappasswd
New password: 
Re-enter new password: 
{SSHA}MlyFaZxG6YIQ0d/Vw6fIGhAXZiaogk0G
Replace 'secret' with the entire hash, {SSHA}MlyFaZxG6YIQ0d/Vw6fIGhAXZiaogk0G, for the rootpw line. Now, let's create a basic schema for my network.
# cat /etc/openldap/schema/hillvalley.ldif
dn: dc=buford,dc=hillvalley
objectClass: dcObject
objectClass: organization
o: bufford.hillvalley
dc: buford

dn: ou=groups,dc=buford,dc=hillvalley
objectCLass: top
objectClass: organizationalunit
ou: groups

dn: ou=users,dc=buford,dc=hillvalley
objectClass: top
objectClass: organizationalunit
ou: users

dn: cn=world,ou=groups,dc=buford,dc=hillvalley
objectClass: top
objectClass: posixGroup
cn: world
gidNumber: 1001

dn: uid=paulie,ou=users,dc=buford,dc=hillvalley
objectClass: top
objectClass: account
objectClass: posixAccount
objectClass: shadowAccount
cn: Paul Johnson
uid: paulie
uidNumber: 1001
gidNumber: 1001
homeDirectory: /paulie/
loginShell: /usr/bin/bash
userPassword: secret
I've created a single group, world, and a single user, paulie. Both share the uid and gid of 1001. LDAP supports lots of additional variables for configuring a user and group account, but I've kept it basic in this example. Once again, be sure to change the domain components to match your network. Feel free to also change the user and group details. I've left the userPassword field in clear-text as 'secret'. The same slappasswd method above applies here as well. It's time to turn on the server, but first, let's change some ownership permissions:
[paulie@buford ~]$ sudo chown -R openldap:openldap /var/openldap/
... and now ...
[paulie@buford ~]$ sudo svcadm enable ldap/server
Check that it worked:
[paulie@buford ~]$ svcs | grep ldap
online         12:13:49 svc:/network/ldap/server:openldap_24
Neat, now let's add our schema file to the database:
[paulie@buford ~]$ ldapadd -D "cn=admin,dc=buford,dc=hillvalley" -f /etc/openldap/schema/hillvalley.ldif
Enter bind password: 
adding new entry dc=buford,dc=hillvalley
adding new entry ou=groups,dc=buford,dc=hillvalley
adding new entry ou=users,dc=buford,dc=hillvalley
adding new entry cn=world,ou=groups,dc=buford,dc=hillvalley
adding new entry uid=paulie,ou=users,dc=buford,dc=hillvalley
That's it! Our LDAP server is up, populated, and ready to authenticate against.

Configuring the Client
I'm going to turn my example server, buford.hillvalley, into an LDAP client as well. To do this, we need to run the `ldapclient` command to map our new user and group data:
[paulie@buford ~]$ ldapclient manual \
-a credentialLevel=proxy \
-a authenticationMethod=simple \
-a defaultSearchBase=dc=buford,dc=hillvalley \
-a domainName=buford.hillvalley \
-a defaultServerList=192.168.1.103 \
-a proxyDN=cn=admin,dc=buford,dc=hillvalley \
-a proxyPassword=secret \
-a attributeMap=group:gidnumber=gidNumber \
-a attributeMap=passwd:gidnumber=gidNumber \
-a attributeMap=passwd:uidnumber=uidNumber \
-a attributeMap=passwd:homedirectory=homeDirectory \
-a attributeMap=passwd:loginshell=loginShell \
-a attributeMap=shadow:userpassword=userPassword \
-a objectClassMap=group:posixGroup=posixgroup \
-a objectClassMap=passwd:posixAccount=posixaccount \
-a objectClassMap=shadow:shadowAccount=posixaccount \
-a serviceSearchDescriptor=passwd:ou=users,dc=buford,dc=hillvalley \
-a serviceSearchDescriptor=group:ou=groups,dc=buford,dc=hillvalley \
-a serviceSearchDescriptor=shadow:ou=users,dc=buford,dc=hillvalley
As usual, change the host and domain names as well as the IP address held in defaultServerList and the proxyPassword. The command should respond back that the system was configured properly, however, additional changes will need to be made if you use DNS for hostname lookups (most people use DNS, so run these commands).
svccfg -s name-service/switch setprop config/host = astring: \"files dns ldap\"
svccfg -s name-service/switch:default refresh
svcadm restart name-service/cache
Now, we need to change how users login so that the client knows that there is an extra LDAP server to authenticate against. This should not lockout local worries. Examine the two files /etc/pam.d/login and /etc/pam.d/other. Change any instance of
auth required            pam_unix_auth.so.1
to
auth binding            pam_unix_auth.so.1 server_policy
After this line, add the following new line:
auth required           pam_ldap.so.1
That's it! Finally, reboot your system and see if you can login with your newly created user.

Update: Glenn Faden wrote an excellent guide to configuring OpenLDAP using the native Solaris user/group/role management system.

Monday Feb 11, 2013

Recovering Passwords in Solaris 11

About once a year, I'll find a way to lock myself out of a Solaris system. Here's how to get out of this scenario. You'll need a Solaris 11 Live CD or Live USB stick.
  • Boot up from the Live CD/USB
  • Select the 'Text Console' option from the GRUB menu
  • Login to the solaris console using the username/password of jack/jack
  • Switch to root
  • $ sudo su
    password jack
    
  • Mount the solaris boot environment in a temporary directory
  • # beadm mount solaris /a
    
  • Edit the shadow file
  • # vi /a/etc/shadow
    
  • Find your username and remove the password hash
  • Convert
    username:iEwei23SamPleHashonf0981:15746::::::17216
    to
    username::15746::::::17216
    
  • Allow empty passwords at login
  • $ vi /a/etc/default/login
    Switch this line
    PASSREQ=YES
    to
    PASSREQ=NO
    
  • Update the boot archive
  • # bootadm update-archive -R /a
    
  • Reboot and remove the Live CD/USB from system
  • # reboot
    
    If prompted for a password, hit return since this has now been blanked.

Monday Sep 24, 2012

ZFS Storage at Oracle OpenWorld 2012

Join my colleagues and myself at this year's Oracle OpenWorld. We'll be hosting a hands-on lab, demonstrating the ZFS Storage Appliance and its analytics features.

HOL10034 - Managing Storage in the Cloud 

October 1st (Monday) 3:15 PM - Marriott Marquis - Salon 14/15
October 2nd (Tuesday) 5:00 PM - Marriott Marquis - Salon 14/15

Monday Feb 20, 2012

CIFS Sharing on Solaris 11

Things have changed since Solaris 10 (and Solaris 11 Express too!) on how to properly set up a CIFS server on your Solaris 11 machine so that Windows clients can access files. There's some documentation on the changes here, but let me share the full instructions from beginning to end.
hostname: adrenaline
username: paulie
poolname: pool
mountpnt: /pool
share: mysharename
  • Install SMB server package
  • [paulie@adrenaline ~]$ sudo pkg install service/file-system/smb
    
  • Create the name of the share
  • [paulie@adrenaline ~]$ sudo zfs set share=name=mysharename,path=/pool,prot=smb pool
    
  • Turn on sharing using zfs
  • [paulie@adrenaline ~]$ sudo zfs set sharesmb=on pool
    
  • Turn on your smb server
  • [paulie@adrenaline ~]$ sudo svcadm enable -r smb/server
    
  • Check that the share is active
  • [paulie@adrenaline ~]$ sudo smbadm show-shares adrenaline
    Enter password: 
    c$                  Default Share
    IPC$                Remote IPC
    mysharename           
    3 shares (total=3, read=3)
    
  • Enable an existing UNIX user for CIFS sharing (you may have to reset the password again eg.`passwd paulie` )
  • [paulie@adrenaline ~]$ sudo smbadm enable-user paulie
    
  • Edit pam to allow for smb authentication (add line to end of file)
  • Solaris 11 GA only:
    [paulie@adrenaline ~]$ vi /etc/pam.conf
    
    other   password required       pam_smb_passwd.so.1 nowarn
    
    Solaris 11 U1 or later:
    [paulie@adrenaline ~]$ vi /etc/pam.d/other
    
    password required       pam_smb_passwd.so.1 nowarn
    
  • Try to mount the share on your Windows machine
  • \\adrenaline\mysharename
    

Tuesday Jan 03, 2012

NFS mounts with ZFS

I ran into a strange automount issue where my NFS shares were not being mounted at boot time. nfs/client was enabled, my entry in /etc/vfstab was correct, and issuing a `mount -a` worked flawlessly. So what was the problem? Well, this was the entry in my vfstab file:
biff:/paulie    -       /export/home/paulie/biff   nfs     -       yes     proto=tcp,vers=3
I wanted to place my NFS share inside a zfs filesystem so that it was easily accessible in my home directory.
[paulie@doc ~]$ zfs list | grep export/home/paulie
rpool/export/home/paulie  78.3M  2.82G  78.3M  /export/home/paulie
Turns out this is not such a good idea since the /etc/vfstab file is read *before* zpool's are imported and mounted. This means that all NFS shares need to be listed outside any filesystems to be mounted at boot time and then symlinked in.
[root@doc ~]# mkdir /biff
[paulie@doc ~]$ ln -s /biff/ /export/home/paulie/biff/
... and then changing around vfstab ...
biff:/paulie    -       /biff   nfs     -       yes     proto=tcp,vers=3
And that's it, NFS should automount now:
[paulie@doc ~]$ df -kh | grep biff
biff:/paulie           2.7T   1.2T       1.4T    47%    /biff
Lesson learned.

Wednesday Oct 12, 2011

ZFS Storage Appliance at Storage Networking World Fall 2011

I was an instructor at SNW this year at the JW Marriott hotel in Orlando, Florida. Along with fellow Oracle co-worker Ray Clarke, we represented the ZFS Storage Appliance in a hands-on environment that allowed storage administrators and industry experts to demo our product in a simulated environment.


Rather than haul physical equipment to the convention, we setup an array of Windows 7 virtual machine sessions paired with a ZFS simulator running on Virtual Box across two remote X4270 machines. This let us create a classroom environment of 24 stations (48 VM sessions) that created a superb replica of the 7000 product that each user could toy around with as they completed the storage exercises we devised.

If you missed the opportunity to demo our product, or would like to download and play with the simulator in your own environment, feel free to check out the following links to get started.

Tuesday May 31, 2011

Compiling Alpine on Solaris 11

I use alpine as my primary e-mail client. In order to get it compiled for Solaris 11 (snv_166 and later), you will need to make a few changes to the source.
[paulie@adrenaline ~]$ uname -orv
5.11 snv_166 Solaris
[paulie@adrenaline ~]$ ./configure --with-ssl-include-dir=/usr/include/openssl
[paulie@adrenaline ~]$ gmake
We run into a problem ...
In file included from osdep.c:66:
scandir.c: In function `Scandir':
scandir.c:45: error: structure has no member named `dd_fd'
Let's investigate:
[paulie@adrenaline ~]$ vi /usr/include/dirent.h
#if defined(__USE_LEGACY_PROTOTYPES__)
/* traditional SVR4 definition */
typedef struct {
        int     dd_fd;          /* file descriptor */
        int     dd_loc;         /* offset in block */
        int     dd_size;        /* amount of valid data */
        char    *dd_buf;        /* directory block */
} DIR;                          /* stream data from opendir() */
#else
/* default definition (POSIX conformant) */
typedef struct {
        int     d_fd;           /* file descriptor */
        int     d_loc;          /* offset in block */
        int     d_size;         /* amount of valid data */
        char    *d_buf;         /* directory block */
} DIR;                          /* stream data from opendir() */
#endif  /* __USE_LEGACY_PROTOTYPES__ */
Interesting, so alpine *should* be using POSIX instead of the older UNIX SVR4 definitions. Let's make a change to the scandir.c file, which is located in alpine-2.00/imap/c-client/scandir.c. On line 45 I see the following use of dd_fd:
  if ((!dirp) || (fstat (dirp->dd_fd,&stb) < 0)) return -1;
Let's change that dd_fd to d_fd.
  if ((!dirp) || (fstat (dirp->d_fd,&stb) < 0)) return -1;
After recompile, everything works as expected. I'm sure there is a better way of fixing this problem, but considering how trivial this issue is, a small edit is sufficient.

Monday Jan 03, 2011

ZFS Encryption for USB sticks on Solaris 11 Express

USB memory sticks are easily lost, so to keep your data safe, it's best to use the new encryption feature of ZFS available since snv_149 (ZFS version 30). Here's how to take advantage of it.
[paulie@adrenaline ~]$ uname -a
SunOS adrenaline 5.11 snv_155 i86pc i386 i86pc Solaris
Get the device id for the USB stick using rmformat.
[paulie@adrenaline ~]$ rmformat
Looking for devices...
     1. Logical Node: /dev/rdsk/c11t0d0p0
        Physical Node: /pci@0,0/pci108e,534a@2/hub@4/storage@1/disk@0,0
        Connected Device: SanDisk  U3 Cruzer Micro  8.02
        Device Type: Removable
	Bus: USB
	Size: 1.9 GB
	Label: 
	Access permissions: Medium is not write protected.
The device id is c11t0d0p0. Using this id, we can make a pool on the device called 'secret'. You can call yours whatever you want.
[paulie@adrenaline ~]# zpool create -O encryption=on secret c11t0d0p0
Enter passphrase for 'secret': 
Enter again: 
Let's create a random 128MB file in the new pool called file.enc.
[paulie@adrenaline ~]# cd /secret; mkfile 128m file.enc
Now, let's make sure it works by exporting and importing the secret pool and hope it asks for a password.
[paulie@adrenaline ~]# zpool export secret
[paulie@adrenaline ~]# zpool import secret
Enter passphrase for 'secret': 
It works as expected. Let's check for the created file.
[paulie@adrenaline ~]# ls /secret
file.enc
We can also check the encryption of any zfs filesystem by using the following command:
[paulie@adrenaline ~]# zfs get encryption secret
NAME    PROPERTY    VALUE        SOURCE
secret  encryption  on           local
For more information visit:
http://docs.sun.com/app/docs/doc/821-1448/gkkih

Friday Jun 11, 2010

Flash Player 10.1 RC available for Solaris / OpenSolaris

Adobe Flash Player 10.1 RC is now available as a prerelease. This version does not require the medialib dependency and should work on its own after installation. You can find it at the Adobe Labs site.

x86
$ wget http://download.macromedia.com/pub/labs/flashplayer10/flashplayer10_1_rc1_solaris_x86_061010.tar.bz2
$ bzip2 -d flashplayer10_1_rc1_solaris_x86_061010.tar.bz2
$ tar xvf flashplayer10_1_rc1_solaris_x86_061010.tar
$ mkdir -p ~/.mozilla/plugins/
$ mv flash_player_solaris_10_1_53_64_x86/libflashplayer.so ~/.mozilla/plugins/
  OR
$ pfexec mv flash_player_solaris_10_1_53_64_x86/libflashplayer.so /usr/lib/firefox/plugins/
SPARC
$ wget http://download.macromedia.com/pub/labs/flashplayer10/flashplayer10_1_rc1_solaris_sparc_061010.tar.bz2
Follow x86 directions above

Wednesday Jun 09, 2010

"Cannot open device" Error

On occasion if you are trying to fdisk or mount a USB disk on Solaris 10 you may get an error that says "Cannot open device." For example:
# fdisk /dev/rdsk/c2t0d0s2
Cannot open device
Things to check:
  • 1. Make sure you are root or have the correct user privilege
  • 2. Try `devfsadm -Cv` to remove any stale disk entries
  • 3. Stop volume management `svcadm disable volfs`

Wednesday Apr 28, 2010

Packages Renamed in OpenSolaris

If you have been using the package management system in the newest builds of OpenSolaris (b133+), you may have noticed the naming scheme changes that affect new and existing packages. [Full list of changes]

For example, trying to find the usb header file package is a bit trickier. These files, namely usba.h, usbai.h, and usbdevs.h, are not installed into /usr/include/sys/usb in the default installation. To retrieve them, use this command:

$ pfexec pkg install header-usb

In previous version, these files could have been obtained from either SUNWusbu or SUNWsfwhea. The removal of the SUNW prefix is the most apparent name change for the package collection. Now you can get busy building the latest apcupsd version!

About

Hiya, my name is Paul Johnson and I'm a software engineer working on the ZFS storage appliance .

Search

Categories
  • Oracle
Archives
« April 2014
SunMonTueWedThuFriSat
  
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
   
       
Today