From the abstract:

We report on our experience with the hardware transactional memory (HTM) feature of two revisions of a prototype multicore processor. Our experience includes a number of promising results using HTM to improve performance in a variety of contexts, and also identifies some ways in which the feature could be improved to make it even better. We give detailed accounts of our experiences, sharing techniques we used to achieve the results we have, as well as describing challenges we faced in doing so. This technical report expands on our ASPLOS paper [9], providing more detail and reporting on additional work conducted since that paper was written.

Anyone who’s interested in High Performance Computing (HPC) or performance gains from Transactional Memory should have a read through this paper - it’s interesting stuff.

Browse Related Posts

• New HPC for Dummies book announced
• Sun’s July HPC newsletter is out
• Sun announces new HPC gear at SC08 trade show
• Mad Larry seeks to rule all with Sun/Oracle
• New SPARC solutions are on their way

Luckily Solaris comes with a handy utility called logadm, which is used by the OE to manage some of the core system log files. logadm can quickly and easily be used to handle all of our log file managment needs.

Let’s look at two log files which aren’t handled by Solaris out of the box - sulog and wtmpx. Both are important, as they help us with our user access audit trail. For starters, we want to keep two old copies of each, and we want to cycle them every two weeks.

Our logadm syntax looks like this:

/usr/sbin/logadm -C 2 -p 2w -c -w <full_path_to_logfile>

• -C number of copies to keep
• -p time between each log cycle (2 weeks)
• -c copy and them truncate (doesn’t need to restart a service then)
• -w writes an entry into /etc/logadm.conf for this log file

So executing the following:

bash-3.00# logadm -C 2 -p 2w -c -w /var/adm/sulog
bash-3.00# logadm -C 2 -p 2w -c -w /var/adm/wtmpx

will result in the following two lines being written to the end of /etc/logadm.conf

/var/adm/wtmpx -C 2 -c -p 2w
/var/adm/sulog -C 2 -c -p 2w

Getting logadm to add an entry to /etc/logadm.conf means that this won’t be a one-off thing - each time logadm executes from cron, it will read the entries from this file. Each entry is checked to see if the log file’s size or age means it’s due for rotation.

wtmpx is obviously a binary file that’s used by last - rather than having to restart the utmpd daemon, it’s easier to just truncate the file. last can still read the older copies - just use the syntax

last -f <wtmpx_file>

It’s important to properly cycle wtmpx, rather than just deleting or truncating it, because it provides a helpful audit trail of users who accessed the system - showing when they logged in, and from where.

This is great if we just want to cycle logs around - but what if we want to compress them as well? Apache is the poster child for log generation - it spits out copious amounts of data, and you want to keep it all for analysis, but it’s a pain to manage.

On my test machine I’ve deployed Apache via Blastwave, so it’s logging to /opt/csw/apache2/var

With SSL enabled there are five log files that I’m interested in:

bash-3.00# ls -l
total 31536
-rw-r--r--   1 root     other    5400214 Oct 30 16:26 access_log
-rw-r--r--   1 root     other    8716843 Oct 29 23:18 error_log
-rw-r--r--   1 root     root      760298 Oct 30 16:26 ssl_access_log
-rw-r--r--   1 root     root      268873 Oct 30 16:26 ssl_error_log
-rw-r--r--   1 root     root      934541 Oct 30 16:26 ssl_request_log

We can just modify the previous logadm command to handle wtmpx - but what about the compression? Helpfully logadm will automatically compress cycled log files using gzip, if we pass it the -z flag. -z will also take a count option, which tells logadm to leave the most recent logfiles uncompressed.

In this case, however, we want everything except the current in-flight log file compressed, and we want to cycle when the logfile reaches 10mb is size:

bash-3.00# logadm -C 10 -s 10m -c -z 0 -w /opt/csw/apache2/var/log/access_log

logadm drops an entry into /etc/logadm.conf for us:

/opt/csw/apache2/var/log/access_log -C 10 -c -s 10m -z 0

Add an entry for each of the five log files, and we end up with this in /etc/logadm.conf:

/opt/csw/apache2/var/log/access_log -C 10 -c -s 10m -z 0
/opt/csw/apache2/var/log/error_log -C 10 -c -s 10m -z 0
/opt/csw/apache2/var/log/ssl_access_log -C 10 -c -s 10m -z 0
/opt/csw/apache2/var/log/ssl_error_log -C 10 -c -s 10m -z 0
/opt/csw/apache2/var/log/ssl_request_log -C 10 -c -s 10m -z 0

Using a combination of log cycling and compression, logadm can handle pretty much any application’s log files for us. By using copy and truncate as well, we aren’t forced to restart each application when we cycle the logs, which ends up giving us a huge amount of control over our log files, without having to write and maintain shell scripts.

Browse Related Posts

• Solaris basics: checking which package a file belongs to
• Solaris basics: configuring the DHCP client
• Solaris basics: configuring ethernet interfaces
• Capturing core files in Red Hat Enterprise Linux
• Empty config file: ‘/var/ldap/ldap_client_file’

Along with the usual Solaris package management commands of pkgadd and pkgrm, there’s a lesser known utility called pkgchk. pkgchk allows us to check which package a file belongs to.

pkgchk will work with binaries:

-bash-3.00$ /usr/sbin/pkgchk -l -p /usr/bin/less
NOTE: Couldn't lock the package database.
Pathname: /usr/bin/less
Type: regular file
Expected mode: 0555
Expected owner: root
Expected group: bin
Expected file size (bytes): 117760
Expected sum(1) of contents: 20724
Expected last modification: Jan 23 01:48:30 2005
Referenced by the following packages:
        SUNWless
Current status: installed

And we can also invoke pkgchk for libraries as well:

-bash-3.00$ /usr/sbin/pkgchk -l -p /lib/libresolv.so.1
NOTE: Couldn't lock the package database.
Pathname: /lib/libresolv.so.1
Type: regular file
Expected mode: 0755
Expected owner: root
Expected group: bin
Expected file size (bytes): 48368
Expected sum(1) of contents: 5063
Expected last modification: Jan 23 01:44:54 2005
Referenced by the following packages:
        SUNWcslr
Current status: installed

For a quick overview of some of the other options, just invoke pkgchk with the -? command line:

 -bash-3.00$ /usr/sbin/pkgchk -?
usage:
        pkgchk [-l|vqacnxf] [-R rootdir] [-p path[, ...] | -P path[, ...]]
                [-i file] [options]
        pkgchk -d device [-f][-l|v] [-p path[, ...] | -P path[, ...]]
                [-V ...] [-M] [-i file] [-Y category[, ...] | pkginst [...]]
        where options may include ONE of the following:
                -m pkgmap [-e envfile]
                pkginst [...]
                -Y category[, ...]

As you can see from the output, pkgchk is also very handy to see if the binaries or libraries that were part of a Solaris package file have been overwritten. As each file in a package has it’s checksum, file size, ownership, and permissions stored as part of the package manifest, this gets added to the Solaris package database when the package is installed - giving a quick and easy method to sanity check your installation.

Browse Related Posts

• Easy Solaris log file management with logadm
• Solaris basics: configuring the DHCP client
• Solaris basics: configuring ethernet interfaces
• Capturing core files in Red Hat Enterprise Linux
• Empty config file: ‘/var/ldap/ldap_client_file’

Instead of a hostname.<interface> file, you need to create two empty files – dhcp.<interface> and hostname.<interface>, in the /etc/directory

bash-3.2$ touch /etc/dhcp.e1000g0
bash-3.2$ touch /etc/hostname.e1000g0

Reboot for the changes to take effect.

NOTE: you can have either DHCP or a static IP address – so double check which files are in place under /etc and what their contents are.

DHCP can be checked with the following commands:

bash-3.2$ /usr/sbin/ifconfig e1000g0 dhcp status

And you can release the DHCP lease with:

bash-3.2$ /usr/sbin/ifconfig e1000g0 dhcp release

DHCP from within VMWare will allocate a default router and DNS settings automatically to your Solaris VM. To enable this to work properly, both /etc/defaultrouter and /etc/resolv.conf must be present, but empty.

If these files already exist, just delete them:

bash-3.2$ rm /etc/defaultrouter
bash-3.2$ rm /etc/resolv.conf

Then recreate them as empty files with the touch command:

bash-3.2$ touch /etc/defaultrouter
bash-3.2$ touch /etc/resolv.conf

Now reboot and note the new settings take effect.

Browse Related Posts

• Solaris basics: configuring ethernet interfaces
• Easy Solaris log file management with logadm
• Solaris basics: checking which package a file belongs to
• Capturing core files in Red Hat Enterprise Linux
• Empty config file: ‘/var/ldap/ldap_client_file’

Each LAN interface in Solaris has a configuration file in /etc which is named after the interface name and it’s instance.

So, let’s assume that you have two e1000g interfaces configured – you’d have two configuration files:

/etc/hostname.e1000g0
/etc/hostname.e1000g1

These files will just contain a hostname, which maps to an entry in /etc/hosts.

An example from my OpenSolaris development workstation:

bash-3.2$ cat /etc/hostname.e1000g0
grond

And this is /etc/hosts:

# Internet host table
#
::1 grond grond.local localhost loghost
127.0.0.1 localhost
192.168.13.100	grond grond.local loghost

When Solaris boots, all it does to configure a LAN interface is look for hostname.* files in /etc. It then looks up the hostname in /etc/hosts to find out the IP address, then configures the interface.

So if you need to change the IP address of a LAN interface in Solaris, all you need to do is edit the hostname entry in /etc/hosts and reboot. For example, to change grond’s e1000g0 IP address to a different subnet, /etc/hosts would be modified to:

# Internet host table
#
::1 grond grond.local localhost loghost
127.0.0.1 localhost
192.168.125.100	grond grond.local loghost

Then reboot the host for the changes to take effect.

Browse Related Posts

• Solaris basics: configuring the DHCP client
• Easy Solaris log file management with logadm
• Solaris basics: checking which package a file belongs to
• Capturing core files in Red Hat Enterprise Linux
• Empty config file: ‘/var/ldap/ldap_client_file’

Key takeaway points from his speech were:

“I would like us to be the successor to IBM”

“We think with the combination of Sun technology and Oracle technology we can succeed and beat IBM - that’s our goal.”

“We are keeping everything. We’re keeping tape. We’re keeping storage. We’re keeping x86 technology and SPARC technology - and we’re going to increase the investment in it.”

“Sun has fantastic technology. We think it’s got great microprocessor technology - it needs a little more investment, but we think it can be extremely competitive. It’s got the leading tape archival systems. We think the Open Storage on their new disk system is absolutely fantastic. Java speaks for itself. Solaris is overwhelmingly the best open-systems operating system on the planet.”

“I shall buy Sun for *seven* *billion* *dollars* muahahahhaha”

“I shall call [Jonathan Schwartz] Mini Me”

Wait, sorry, no - those last two I made up. Although I’m sure Larry is enjoying a Bond villain style evil laugh as he contemplates finally putting the boot in to IBM, whilst stroking his devil goatee.

Joking aside, he’s right - Sun has a lot of really excellent technology, and historically they’ve failed to follow through with most of it. Oracle’s ruthless sales methodology will be striking some fear into some Sun sales managers, but engineers (and customers) should be taking comfort hearing this sort of fighting talk.

Of course, the deal (still!) isn’t done, and it might be a while before we see some systems hitting the market on the back of this enthusiasm. Still, I think Sun have totally neglected the higher end systems, relying on Fujitsu to do the hard graft - with any luck, we should be seeing some big Niagra boxes coming soon.

Browse Related Posts

• New SPARC solutions are on their way
• Larry Ellison on Cloud Computing
• Oracle have bought Sun
• Sun’s research report on Hardware Transactional Memory
• New HPC for Dummies book announced

Couple this with the blurring between ‘traditional’ HPC systems for research, and new high end business solutions for statistical analysis and database warehouses, and there’s a real need to de-mystify HPC for all.

Douglas Eadline, working with Sun and AMD, has written HPC for Dummies, and the blurb explains:

This special edition eBook from Sun and AMD shares details on real-world uses of HPC, explains the different types of HPC, guides you on how to choose between different suppliers, and provides benchmarks and guidelines you can use to get your system up and running.

What makes this really great is that this is a free ebook, available direct for download from Sun’s site. I can highly recommend that anyone with an interest in HPC (or just large scale systems design) grabs this and has a read through.

Head on over to Sun’s HPC for Dummies page.

Browse Related Posts

• Sun’s research report on Hardware Transactional Memory
• Sun’s July HPC newsletter is out
• Sun announces new HPC gear at SC08 trade show
• Mad Larry seeks to rule all with Sun/Oracle
• New SPARC solutions are on their way

However, Sun also offer two extra repositories – one for those with a support contract for OpenSolaris, and one containing extra software – a bit like the old Extras CD that came with the Solaris media pack.

The main site for accessing all of this is https://pkg.sun.com If you don’t already have a Sun ID and password, you’ll need to follow the links from that page to register for them (don’t worry, it’s free). Then login, and you can see you’re presented with links to generate Certificates for the two repositories. Let’s assume you’re not a corporate user and don’t have a support contract, and concentrate on the Extras repository.

Click the link to generate the key and certificate for pkg.sun.com/opensolaris/extra and save them somewhere safe – you’ll be needing them again. For example, I tend to stick them in /var/ssl/pkg as well as keeping copies in my home directory.

Now what we need to do is add the Extras repository as a publisher in the package management system, and then link the key and certificate to that publisher.

So in this example we can accomplish this in one command (note the backslash to fit it across two lines):

$ pfexec pkg set-publisher -O https://pkg.sun.com/opensolaris/extra \
-k /var/pkg/ssl/OpenSolaris_extras.key.pem -c /var/pkg/ssl/OpenSolaris_extras.certificate.pem extra

All this does is:

• adds a new repository at https://pkg.sun.com/opensolaris/extra
• links the key (-k) /var/pkg/ssl/OpenSolaris_extras.key.pem to it
• links the certificate (-c) /var/pkg/ssl/OpenSolaris_extras.certificate.pem to it
• gives it the name ‘extra’ in the package manager

That’s all there is to it. Now you can use the standard OpenSolaris package management commands:

• pkg refresh will update the catalog with the packages from the new repository
• pkg install can be used to install them

And of course, if you’re using the OpenSolaris package manager gui, you can select the new repository from the drop-down list of available repositories, and then browse through from there.

Browse Related Posts

• A quick and easy start to HPC development
• A review of the Sun Ultra 24 workstation
• Sun’s Unified Storage Simulator

Boasting 16 cores, each with it’s own cache, it’s an impressive bit of silicon. Each core has 4 Coherency Units (COU). Familiar to anyone who’s played with big Silicon Graphics kit and other cc:NUMA boxes, Coherency Units keep track of memory contents (from L2 cache up to physical RAM).

4 COUs per core, along with 16 seperate L2 caches, is a lot on a single die. To simplify things (and to help improve performance) Sun has added a Core to Cache Crossbar (CCX). Two cores will share a single entry point into the CCX, which is linked to every core on the chip.

It’s essentially a standard switching crossbar - like UPA, Xbow, NUMAlink etc. - except it’s linking cores and L2 caches, not CPUs, RAM, and I/O bays.

All pretty impressive stuff - remember, this is all on the processor die - and it sounds like it would be ideal in a large system. Those M9000 boxes look pretty inviting, especially after the disappointment of the Niagara system board upgrades for the F15k.

And in fact, maybe that’s what Oracle has up it’s sleeve. The Prophets of Larry have said they will be making a major SPARC announcement on October 14th, during Larry’s yearly sermon to the faithful at Oracleworld. (Sorry, guys - I love you really)

With IBM pushing forward development of it’s fearsome Power7 chip, and with Power6 a bit of a monster anyway, Sun clearly needs to keep in the game - especially after (foolishly) canning Rock.

The Oracle announcement will be aimed clearly at blowing the doors off IBM - check their rhetoric and the fuzzy teaser advert on the Sun plus Oracle is Faster page.

Now, there are a number of ways they could bury IBM in the TPC-C benchmarks. Most obvious would be a massive RAC install, probably with Fujitsu’s new 8 way SPARC64 chip. But that wouldn’t really be blowing the Sun SPARC trumpet, would it?

The current top end Niagara box - the T5440 - can have up to four 8 core Niagara CPUs (along with half a terabyte of RAM). It’s pretty good, but again - the only way you can scale is by clustering them. Oracle have an insanely great scalable database solution with Oracle RAC, so it would seem a no-brainer.

However, if Sun are close to releasing Rainbow Falls, they could use the system boards and interconnects from the M9000 chassis to produce a hugely thread-dense NUMA machine. All that coherency hardware makes no sense for a cluster - it’s role is in a big Single System Image (SSI) machine.

We’ll find out in a month.

Browse Related Posts

• Mad Larry seeks to rule all with Sun/Oracle
• Sun’s OpenSPARC CPU
• Larry Ellison on Cloud Computing
• Oracle have bought Sun
• Sun’s research report on Hardware Transactional Memory

Have a look through the initial announcement, download OpenDS 2.0 then grab the release notes and have a quick look through the documentation.

As always with OpenDS the install is very straightforward and painless. Once up and running the GUI makes a nice change from the slow, cluttered mess of the old Directory Server. I’d now rate OpenDS is a solid, mature offering, and I’ll be looking at replacing some of my aging DS 5.2 installs with it.

OpenDS qualifies for support from Sun, in it’s Sun OpenDS Standard Edition guise, so there’s no reason not to look at production deployments. For the more esoteric installs, you’ll still want Sun Directory Server (v.7 is being working on at the moment) but for a smaller or more straightforward install, OpenDS will fit the bill nicely.

Browse Related Posts

• Modifying LDAP entries in Sun’s Directory Server
• Sun’s research report on Hardware Transactional Memory
• Mad Larry seeks to rule all with Sun/Oracle
• New HPC for Dummies book announced
• New SPARC solutions are on their way