Busy day out of here!
So! Morning soon here to re-start the computer (yes, again) and put the SDSS downloads to work. It's 1/5th done! And also finished the PSF and put GALAPAGOS to run in one more cluster.
Next step is prepare the images for the 2 last clusters and also do the multi-component 2D fit to our observed galaxies.
Then the doctor appointment and it's all fine with Matheus, he's in the right position to arrive, that means it can be anytime now ...
After that downtown to final Xmas shopping! DONE!
Back here to check how things are going and a good talk by phone with the boss about good thing, which are still not publishable! :-)))
Also yesterday and today will keep going, reading the final version of my friend Miguel Verdugo's paper! It's good so far :-))
It keeps going ... :-)
The download of SDSS goes. The day was pretty much downloading the files, the net was VERY slow today, don't know if here or at the hosting server ... anyway, now it's going to the proper disk.
From science the preparation of the images to GALAPAGOS analysis. Flagging bad objects and so on. Now gotta do the PSF and then the GALFIT part runs. Tomorrow PSF, hopefully in the morning. We have doctor appointment for Matheus check up, but I hope to do something before it yet!
And they keep happening!
Yesterday was a normal day. The computer crashed a couple of times and I decided to go "safe" and do one thing at once. Like, unzip, then copy to the external disk, then checksum, then unzip the next ... otherwise the computer gets "tired" hehehehe I'm about to go safer and do it in my laptop! Of course, even going safe the computer crashed in the night ... blah!
On science. The cluster I was running it done, the one was running in Innsbruck is done and I'm preparing the next images.
Computer problems
Basically a computer problems day!
During the weekend I had several crashes of my machine due to some "overload" trying to download, checksum, unzip and copy to external disk the large SDSS files, "all at once". The best guess is a temperature problem ...
Anyway, that happened on the same time as I found out I couldn't have a file larger than 16 GB in the external disk. After googling I found the reason (a bit tricky because value that ls -lh showed was 17 GB because it shows in base "1000" and not "1024"). Ext2 with a block size of 1K has a limit of 16 GB per file and it increases to 256 GB with a block size of 2K and so on. So, reformating the whole array was the solution.
Then with so many crashes the file system required a manual check, only done by the sysadm, who was only available in the afternoon. So, no desktop almost the whole day. But when that happens my whole account stops, until today for an unknown reason. The other sysadm (that takes care of the servers) found out why. Environment variable defined to a local disk requires a time out on every action. OK, variable redefined and things work.
Also, file system checked and now the array mounts with the correct file system and at the desired mounting points. That should make the manipulation of data in the disks faster (I hope, checking it NOW). Also I'm testing the temperature issue putting a "real" fan blowing at the open cpu. Let's see.
In the science field. Preparation of two more clusters is going on. One is still running and the other one is ready, there in Innsbruck. The question is how to bring the 16 GBs here.
Another troubleshooting day!
So!! (yes, it starts with "so", again ... and again I said that here!).
Did some small things that should have been done for some time already, like the small SPP report on the trip to Innsbruck, calling Amazon on the shelves for Matheus room, that never arrive (actually the add in Amazon.de was wrong and I've got the wrong product ... blah) ...
Also finished the cross table between catalogue and slit number and the rest of the day was spent on trying to get GCC 3.4.6 working locally to be able to compile "sector" to download the SDSS tables. Now it's downloading.
We also had, at the end of the day, our Xmas party in the institute. A calm dinner with all the people, like institute's parties! :-)))
A photo of the "array" (fortunately it doesn't have to be a RAID :-)) ).
Array Ready!
Day of solving problems.
After almost getting crazy with my mac formating disks with ext2 and cursing a lot on that, I managed to understand how to make it do what I want! Good! Anyway, the ext2 support of Mac OS sux. Apple ignores it, so no official support and the only option is an old package that needs some improvements ...
Disks got ready, physically assembled and mounted. Now they're gonna stay on for a while to physically test them.
The problem of the 2GB files seems to be caused by GCC 3.3 (according to SDSS support), so I'll try to get 3.4 installed. And the files don't have to be in a raid like structure where the whole pack is in the same "disk".
IDL x spawned GALFIT is still going on. GALFIT works on our 64 machines, but not when they are spawned in IDL. actually, it works with X11 lib error messages in some of the machines and don't work at all in other machines ... the ball is back at the sysadm field. All that shows up because GALFIT opens an XTERM while running (which in my opinion is not really necessary, it could have a "non XTERM opening mode" - eventually it has and I don't know). BTW there are small differences on the results for the exact same setup using GALFIT for 32 bits (2.0.3) and for 64 bits (2.0.3c), they are small, but they are there!
GALAPAGOS is still running in one of the clusters in the first stage, the one that models the brightest 5% objects. Then I can split it in 4 processors/machines.
I also compared the objects detection in SExtractor 2.5 using the linux version and the Mac OSX version (that comes in SciSoft OSX). There are differences at the faint regime. Some object are detected by one and not by the other and vice-versa.
Mac version gives a different sky subtraction, no idea why, while 2.5 and 2.3.2 for linux give the same sky and the same number of objects ... still to find out why! Already asked Nor Pirzkal but got no reply up to now. Didn't have time to try to compile it myself ...
Not to say it was a 100% technical day, I started the cross-identification of the objects to be observed, their catalogue number and their slit number (not that this is a GREAT task, but it takes a little time, and it has to be done, and it has to be done in linux, since FIMS doesn't work in Mac OS).
Some days without posting!
Friday was a day to prepare the next cluster to run galapagos on it, but I had to stop at the point of preparing the PSF. The idea is gathering as many good stars as possible in all tiles and I had to learn how Marco's task was working.
Monday was quite a useless day with a lot of paperwork, trying to get the IDL problem fixed, check the final mask for the observing run of this week and buying some external disks to make a large array.
Tuesday was putting the PSF routine to work and visiting the hospital where Matheus will be born!!
Today was a day full of small problems.
1- The bureaucratic IDL support in Germany "CREASO" refused to give any kind of support to my question, because our "license/maintenance contract" is expired.
This fact really made me think is not worth developing a single extra line of code for IDL since their license is so expensive and their way of thinking is so cheap!!
BUT! I wrote the support in Brazil, the SULSOFT and Mr. Michael Steinmayer, who I happen to have met some years ago and he kindly agreed to at least try to help me in this issue.
THANKS A LOT MICHAEL!!! You may solve solve my problem soon, but you surely improved my concept on ITTVIS. There's still hope for IDL.
The problem is that when I SPAWN a GALFIT in any 64 bits machine of our network it complains that it cannot find the X11 library, even though the path is correct, variables are properly set and the file IS THERE. The same setup and file is used in the 32 bits machine and the SPAWN works properly.
If anyone has a clue about it, let me know!
2- Started to format my external disks as ext2 in my MacBook and it collapsed leaving my disk "busy". I had to reformat it in WINDOWS at home to be able to make the MacBook read the disk again ... (it has happened twice). Tomorrow is last try! It can be the disk, since the first external disk format perfectly.
3- Downloading the SDSS SQL files. 24 GB the first one and it quits when the file reaches 2 GB with a "file size limit exceeded" ... even though I can create a file as big as 24 GB "by hand" in this same disk!
Here also, any hint is welcome!!!
Tomorrow back to science!
Parallel curse is over
So! Morning doctor appointment for Matheus check up! Everything is OK and from now on he can come any time!
Lost one of the lessons of the parallel programming curse, but ... :-)) In the afternoon the last lesson. It was a good curse. Of course it requires now more programming and working on that kind of logic and syntax.
Finished the image preparation for the second cluster to be "GALAPAGED". The first one is running the last tile. Tomorrow the second will go on the SExtractor part and object selection, while the last tile finishes.
Parallel programing
So! Morning and afternoon in the Parallel programing course at our computational center (GWDG). In the morning the concepts of parallel processing and some "options". In the afternoon the first class on MPI (Message Passing Interface - not Max Planck Institute hehehehe). It goes on tomorrow.
Later in the afternoon preparing more Observing Blocks with the new mask for the observation this month.
GALAPAGOS is running, one tile is over!
Clumsy day!
During the weekend I left GALAPAGOS running. There were library problems, license problems, but it's running, slower than it could, but running.
Also a last minute call from Bodo to take care of the final stages of the phase 2 of our run this month at VLT. All arranged, but full of "solving things".
Tomorrow the parallel programing course starts. This will be my day for the next 3 days.
I just would like to thank Marco Barden, Thomas Kronberger, Wolfgang Kapferer, Sabine Schindler and all her staff for the pleasant week in Innsbruck.
