blog

Hi there, big projects going on, and not much time to blog theses days, but I hope to fix this soon…
This entry is just to tell you that next week, I should be in Las Vegas (from sunday to friday).
So, if you have the time to chat a little about “anything“, and for coincidence you will be there too, it would be great! Drop me a line, or use the comments section…
peace

103 anos de Glorias!!!

Well, this is that kind of things that could be seen as innovative, and we could have many things to talk positively about (actually we do)… but they really do it the *wrong* way, and so you really think they are not serious. And so, the negative points smash the good ones.
I’m always looking at the Operating Systems new features, as I really don’t know how the administrators handle in these days, some problems like the simple fsck downtime… or the big one: corruption and data loss.
Almost ten years ago, I do remember when we did swtich from extX filesystem on Linux to XFS. Performance, fast on recovery from server crash, and many other features. A filesystem that really helped us to solve many issues. No way to have email workload for example, running on a filesystem that did need fsck, right? It’s insane…
So, I’m looking really close to the new features of Windows 8, and yesterday I was reading this post. And as it was the first good techincal article I found about the underlying design and features of the new filesystem from Microsoft, the first thing I did was “Ctrl + F” to look up for ZFS. And I did find one, on the comments section…
Sorry, I think the ZFS developers are not worried about it, Jeff and Bill should not even know about this new filesystem, but I think Microsoft guys had the obligation to mention it. This new technology from SUN did change the storage Market, and the ZFS creators did talk about some inspiration on the NetApp filesystem and design. There is no shame in doing so…

Why? Because is nice to show some respect.
Microsoft never did it, why they would do it now?

But Microsoft seems to still have the same approuch to make it seems like they did create the Personal Computer, graphical desktop, TCP/IP, and even the browser’s tabs. That’s a shame…

My first intent on this post was to write about some really good feautures this new filesystem seems to have implemented (and my concerns about the NTFS compatibility too). As my experience talking with some SUN/ZFS engineers, the VM was greatly impacted on the implementation of ZFS. And we do not need to talk about the ARC. But…
For now, I will pay some beers for my friends Windows system administrators (If they will have that life saving features, they need to celebrate). And later on, I will try to discuss here on my blog about the positive points ReFS and Storage Spaces will provide to Storage Market and Microsoft world, bringing to them some ZFS features… ;-)
peace

PLEASE BE AWARE THAT ANY INFORMATION YOU MAY FIND HERE MAY BE INACCURATE, AND COULD INCLUDE TECHNICAL INCACCURACIES, TYPOGRAPHICAL ERRORS, AND EVEN SPELLING ERRORS.

DO NOT TRY IT IN PRODUCTION. USE AT YOUR OWN RISK!
For Official Support, please consult http://wwww.isilon.com

For some days I was having a little problem with the generation of the Isilon InsightIQ scheduled reports.
The problem was the resources allocated for the virtual machine appliance, and the GUI was failing miserably… please, if you create an appliance, give some errors, do some indication of what “in the earth” can be the reason for the fact that a user is issuing a command and *nothing* happens!
For sure support says “that was on the documentation for the blah blah blah installation”. Something on the documentation does not give you the right to not display error messages or handle errors on your application.
Actually the provisioning for VM’s on our environment is automated and if we are talking about a “appliance VM”, the configurations should be on the xml file or something. But that is not the case…
After add the minimum memory required for the virtual machine, everything was working fine. But how to create the older reports? We need to generate them… Oh ok, I need to edit the report, change dates execute it and etc? That is some kind of joke right?
Well, I did need to generate ~15 reports (that was the time support took to end this case), and so I could manage to automate the process to create the reports like this:

First step was to figure out the command to create the reports, and I could get some clue from the appliance’s logs (/var/log/insightiq.log).

Second, we need to inform the timestamp to the command to create the report, so we need to convert the actual date to it. Here is just one of the ways you can do it:

 REPORTSTAMP=`date -d 2012-03-12 "+%s"`
 

And here is the actual command to generate at the command line a report on InsightIQ appliance (some variables to make it simpler. Remember to make changes in the cluster_guid and path to your datastore):

WKHTMLTOPDF="/usr/bin/wkhtmltopdf-i386";
WKHTMLOPT="--javascript-delay 110000";
URL="http://localhost/reports/performance/scheduled? \
start_time=$REPORTSTAMP&interval=86400&performance_report_id=user- \
4&cluster_guid=xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx";
FILE="/net/mycluster/ifs/myshare/generated_reports/ \
daily_perf_mycluster_$REPORTSTAMP.pdf";
 

So, the next step is just to execute:

 $WKHTMLTOPDF $WKHTMLOPT $URL $FILE >/dev/null;
 

After that I was thinking that my report would be on the Web GUI and I could visualize it and send it to my email… wrong. The Web GUI pick the reports informations from a SQLITE3 database (I think because users can have *a lot of* saved reports). So, after some digging (and internet search to know some basics about SQLITE3):

# SQLITE3 (Need to update the DB to be able to see the new report on web gui)
# Stay in mind that this is just a quick and dirty way to guess the current id.
# We are not handling cuncorrent executions, or NASDAQ variations...

# Guess the current id from the table...
CURRID=`sqlite3 /net/mycluster/ifs/myshare/archive.generated_reports.sqlite3 \
'select * from generated_reports' | tail -1 | awk -F\| '{print \$1}'`

# Increment the id...
NEXTID=`expr $CURRID + 1`

# Insert the new report on the database (sqlite3)...
sqlite3 /net/mycluster/ifs/myshare/archive.generated_reports.sqlite3 \
"INSERT INTO 'generated_reports' VALUES($NEXTID, 'user-4', \
'Daily Perf mycluster', 'mycluster', '24-hour', $REPORTSTAMP, \
'daily_perf_mycluster_$REPORTSTAMP.pdf');"
 

Now yes, we should have our new report on the Web GUI on InsightIQ, and be able to visualize crucial informations about the cluster performance. I need to say that we have really rich informations on the InsightIQ reports. And the interface is really clean (actually too much, because no errors or warnings are displayed).
I’m learning some neat commands on this platform to solve some problems (like the above one), that support takes days to answer. I will try to post here some of them in case somebody else needs and have the same support experience as I did.
peace

I really like geek T-Shirts! Actually we saw a lot of creativity on it, carpets (who don’t know about the “127.0.0.1″ ;-), cups and etc. I think these kind of stuffs are really cool to spread the word, and that plays a crucial job on Open Source projects.
So, for a long time I had a plan to create some t-shirts for my blog, to promote my book, and try to create some kind of art. ;-) So this time I put the things together and thanks to Linda from the Illumos community, I could find a good place to make this idea come to the real world.
I have many ideas to some graphics and texts from what you can find on my blog, what I really like to write about (technology, music, f1, etc): “Computing Science, Posix rules, Life rules, no rules…“. But I will need some time to add those to my shop (it’s just a hobby), but hope you can enjoy what is already there!
The first one is from a recent idea that I had from the first time I did hear the song from Jay-Z (I really like Music, all genres from hip-hop to classic), and if you don’t know him or do not like Hip Hop music, just take a look at one of my posts called Assorted. I think at least will be funny, and you will understand the joke… ;-)
Another one that could not be forgotten is the ZFS Internals series from my blog that until today has a lot of hits every month!
So the same logo I did create for the PDF (draft) of the series, I did use to promote this section on my blog. And as ZFS plays a big role here too, the “SEND AND RECEIVE” t-shirts to promote Open Source Principle is present on a exclusive section on the shop.
Well, I hope you enjoy and hope I do receive mine real soon! ;-)
peace