Speech:Spring 2015 Melissa Bruno Log


 * Home
 * Semesters
 * Spring 2015
 * Proposal
 * Report
 * Information - General Project Information
 * Experiments - List of speech experiments

Week Ending February 3, 2015
Reading Carnegie Mellon's documentation on Sphinx, speech, and speech recognition.
 * Task:

No expected artifacts.
 * Results:

Read through the "Learn" and "Research" parts of Carnegie Mellon's Sphinx web page. None.
 * Plan:
 * Concerns:

Week Ending February 10, 2015
02/07

We have instructions from Forrest, but it seems like we will not be able to install Redhat on the Metho drone until Tuesday, because Pandora is closed on weekends and will likely be closed on Monday because of the storm. We need to thoroughly document everything we do. A baseline train between Caesar and Asterix needs to be done.

02/08

Durham and the Manchester school district cancelled classes, and UNHM will likely do the same, so I will be going in at 5pm on Tuesday to install Redhat on Methos and document that process. If that is successful, then we can install Redhat on Brutus and run the 3 non-baseline trains.

02/09 The school was closed today, so I will be going in (I think by myself) tomorrow at 5 to get as much of the Redhat installation done as possible. A couple group members will pick up where I left off when they come in on Wednesday. During and after class we'll do the remainder of the work that requires us to be physically present, if any remains.

02/10 Spent 1 hour and 45 minutes in the server room trying to install RedHat on Methusulix. It was clear early on that the instructions we got were not going to work, and every possible combination of pushing buttons and removing the hard drive and changing configurations did not produce any results, so we are going to have to do a much more thorough investigation of why Methusulix isn't reading the disk to install the OS.

Error:

"CLIENT MAC ADDR: 00 0D 56 FD 77 1D GUID: 44454C4C 3500 1042 8054 B7C04F393431 DHCP (times out)

PXE-E53: No boot filename received

PXE-M0F: Exiting Broadco PXE ROM Strike F1 to retry boot, F2 for setup utility"

Error code documentation: http://h20564.www2.hp.com/hpsc/doc/public/display?docId=c04022688 Read through old logs and getting a better sense of what the Systems team needs to accomplish.
 * Task:

Gain a better understanding of what the Systems team needs to accomplish.
 * Results:

Read through old logs. None.
 * Plan:
 * Concerns:

Week Ending February 17, 2015
3/12
 * Task:


 * Asked coworkers about error we started getting from the previous week, and they said it sounded like a DVD drive issue. The installation disk is a DVD, which cannot be read by the CD drive we were trying to use. One person offered to bring in an external DVD drive.

3/15

Set up user accounts on Obelix for the Tools group.

3/16


 * Met with the systems group in the server room with the intention of getting Verleihnix to be accessible to the Internet.
 * After dealing with the power outage and being unable to get work done, we decided to look into the issue individually and reconvene after figuring out the correct way to get Verleihnix to connect to the Internet through Caesar.

3/17

Reasearched network bridges.

3/12
 * Results:

3/15
 * Professor Jonas switched out the CD drive we were using with the one DVD drive on the server rack so that we were able to complete the installation.
 * Revised existing documentation to include more details on the installation, including the trials we had getting to the installation screen.


 * User accounts for the Tools group on Obelix are all set up.
 * I altered the script that Professor Jonas made a bit so that only people who needed access had accounts made for them.
 * Spent a few hours changing the network configurations of Verleihnix remotely to try to get it to connect to the Internet,
 * Arranged to meet with the rest of the Systems group at 3pm in the server room on 3/16.

3/16
 * Met with the systems group in the server room at 3pm.
 * ...The servers were all off and would not turn back on.
 * Through trial and error we figured out that the servers were off because of a power outage, not because they had been changed or were damaged in any way.
 * I talked to Daphne, the security guard, and she offered to reach out to the building's management about the power outage.

3/16 3/12 We couldn't even get to step 1 of the instructions from previous semesters for how to install Red Hat, despite spending considerable amounts of time in the server room trying every way to boot up Caesar so that it would read the installation disk and star the installation.
 * Plan:
 * After dealing with the power outage and being unable to get work done, we decided to look into the issue individually and reconvene after figuring out the correct way to get Verleihnix to connect to the Internet through Caesar.
 * Concerns:

Week Ending February 24, 2015
02/19

Caesar went down momentarily. Went to the UNHM server room and got it back up and running.

3/20
 * Task:
 * Research network configurations, network bridges, and DHCP.

3/23
 * Stayed in the server room working on network configurations from 5 - 8:45pm trying to get Verleihnix on the Internet.
 * Notably found the system-config-utility, which allowed us to turn on and off DHCP
 * Verleihnix successfully connected to the Internet through Caesar

3/23
 * Results:
 * Spent 5pm - 8:45pm trying different methods of getting
 * 10 minutes after I had to leave,

3/22
 * Plan:
 * Agreed to meet with the Systems group in the server room at 5:00pm on 3/23.


 * Concerns:

Week Ending March 3, 2015
3/25
 * Task:
 * Install RedHat on Brutus

3/26
 * Investigate next step of the migration: setting up the file system on /mnt/main and then performing the system migration.

3/25
 * Results:
 * Met with the Systems group after class to install RedHat on Brutus
 * We installed RedHat on Brutus
 * I documented the installation process

3/2
 * Went to server room, tried to change partition sizes so that we could mount the filesystem
 * Tried to change the partition sizes, but none of us could figure out how to do it

3/3 ssh root@caesar "(cd /mnt ; tar cf - . )" | (cd /mnt ; tar xvf -) ps -ax | grep tar
 * Filesystem was mounted thanks to Chris
 * I migrated the file system over using the following tar command while logged into Brutus:
 * What this command does is: SSH into caesar and zip /mnt and all its files into a tar file, then on Brutus unzips them into the /mnt directory.
 * I tested this beforehand with the file /testtest/mnt/main/test and it showed that the permissions were not altered, but permissions and symlinks still need to be verified.
 * Spent the rest of the night into the next day using this command to check if the copy had finished or not:

3/26
 * Plan:
 * Investigate next step of the migration: setting up the file system on /mnt/main


 * Concerns:

Week Ending March 10, 2015
2/4 Directories where I verified permissions, symlinks, and files: /mnt /mnt/main /mnt/main/Exp


 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending March 24, 2015

 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending March 31, 2015
3/26 A student could not create a new directory in Exp 0265, which I fixed by changing the group write permissions with the command chmod g+w.
 * Task:

3/30
 * Miraculix, traubadix, majestix, idefix, automatix, and methusalix needed to be mounted to Caesar
 * new user accounts needed to be created
 * Registration issues also needed to be investigated.
 * ACL implementation feasibility needed to be investigated.

3/29 Investigate changing the default permissions of /mnt/main/Exp to grant write access to group members in newly created directories.

3/31 Perform group tasks logged on group wiki page.

3/26 Successful directory creation with added group write privileges using the following: [mjp238@caesar 0265]$ su - [root@caesar ~]# cd /mnt/main/Exp [root@caesar Exp]# chmod g+w 0265 [root@caesar Exp]# su mjp238 [mjp238@caesar Exp]$ cd 0265 [mjp238@caesar 0265]$ mkdir 001 [mjp238@caesar 0265]$
 * Results:

Error could be replicated by removing group write privilege using the following: [root@caesar Exp]# chmod g-w 0265 [root@caesar Exp]# su mjp238 [mjp238@caesar Exp]$ cd /mnt/main/Exp/0265 [mjp238@caesar 0265]$ mkdir 001 mkdir: cannot create directory `001': Permission denied
 * Plan:

3/30
 * Met with the systems group at 5:30pm.
 * Mohammed and Chris mounted Miraculix, traubadix, majestix, idefix, automatix, and methusalix to Caesar
 * I looked into a concern about registration and Obelix and reset a student's password on Caesar
 * I also made new accounts on miraculix, traubadix, majestix, idefix, automatix, and methusalix.
 * Apparently Mike also made accounts on these servers soon after, and I suspect that Krista's password is not working again because the automated script rejects using her username as a password because it contains three of the same number.
 * We all decided not to make changes to the ACL because it requires unmounting and remounting the file system, and we do not want to disrupt trains. Mike later reached out to us and affirmed this decision.

3/27 The default permissions for newly created directories in the /mnt/main/Exp folder could ordinarily be changed with the following commands: caesar:/mnt/main # chmod g+rwxs Exp caesar:/mnt/main # setfacl -d -m g::rwx Exp

This does not work because by default mounted filesystems have access control list support disabled. Changing /etc/fstab so that it contains the following and then resetting the filesystem will allow the setfacl command to work: /dev/sda /home ext3 defaults,acl 1 2 3/26 Permissions for the Exp directory on both old Caesar (now Brutus) and new Caesar make every new directory lack write privileges for allowed groups. I plan on looking into a fix for this.
 * Concerns:

3/27 Because fixing ACL support requires a system restart or unmounting and remounting the filesystem, this change should be made when the Systems group next meets this week to minimize the number of disruptions to the system and trains that may be running at the time.

3/30 Jonas reached out to the systems group, and because some of the groups are running trains that will take a very, very long time to complete, ACL changes will not be made until after the semester has ended.

Week Ending April 7, 2015
4/01 Going to reach out to Mohammed tomorrow to see if he still wanted to do the recopying of the "old" directory, and resend over the tar command I used. Met with the group today. I'm going to work on task 3 of the 5 tasks Garret shared with the group.
 * Task:


 * Results:


 * Plan:


 * Concerns:

Week Ending April 14, 2015
4/8
 * Task:
 * Met with Patriots group. Me and two other people will be investigating the language model.

4/13
 * Activate RedHat on Obelix
 * Install GCC on Obelix

4/11 Professor Jonas mentioned a while back that he didn't like like how SSH times out so quickly on Caesar, so I intend to find a fix and either implement it or document it for next semester to implement so that the risk of disrupting in-progress trains is zero.

4/8
 * Results:
 * Sent out the poster template and UNH's guidelines for posters at the URC to everyone.
 * Met with the systems group, we came up with the design for our poster together, I grabbed the UNHM, Redhat, and Dell logos to use on the poster.

4/11 It should be possible to stop the timeout issue by adding the following line to /etc/ssh/sshd_config on Caesar: ClientAliveInterval 60

I will create a to-do list for next semester and add this along with the gcc install instructions.

4/13
 * Met with Mohamed in the server room and completed the following tasks:
 * Activate RedHat on Obelix
 * Install GCC on Obelix
 * Documented the above tasks and added that documentation to the Systems page

4/12 Added a section to the Systems Group page of steps that need to be taken by teams in future semesters. Changes in this section can't be made right now due to the risk of disrupting work being done by the current semester's teams.

Sections include:
 * Stop SSH timeouts on Caesar (complete)
 * gcc Installation (instructions not yet done - systems group is meeting on Monday in the server room to do this)
 * ACL (complete)


 * Plan:


 * Concerns:

Week Ending April 21, 2015
4/18 Doing a write up related to my findings on 4/15. Waiting for the private wiki pages to go up before posting, but still going to share a final suggestion for changes to be made with group via email sometime on 4/19.
 * Task:

4/19 Doing more work on my write-up. I'm going to send whatever I have out to the group before class on Wednesday so that we can decide what changes to make after evaluating the effects of the change made last Wednesday.

4/15 Made a great discovery when looking at scripts during class, and Mike confirmed that it seems like a significant find. Communicated findings to group by starting up a new email chain, and started investigating later that night.
 * Results:

4/21 Suggestions writeup finished, just need to edit them and send them out.


 * Plan:


 * Concerns:

Week Ending April 28, 2015
4/22
 * Task:
 * Sent out findings from research done the previous week to the group
 * Met with group after the URC
 * Discussed findings with group
 * Asked Professor Jonas for feedback

4/24
 * Read about speech recognition as a whole on Carnegie Mellon's website

4/25
 * Fought with cisunix.unh.edu for hours. It's not letting me on from my home network and I still don't know why. Nmap scanning cisunix shows that port 22 is closed, but the same scan shows that port 22 is open when I scan from my work network, and Dakota and everyone else haven't had this problem.

4/28
 * Run trains
 * Do more research

4/22
 * Results:
 * We basically have three avenues for making improvements

4/27
 * Ended up downloading the UNH VPN on my personal computer, and now I use this to connect every time I need to get onto Caesar. I still have no idea why I can't just ssh in normally from either of my machines, even though I've had no problem doing so until this week.

4/28
 * Train is running
 * Did some more digging and shared results with group
 * Plan:


 * Concerns:

Week Ending May 5, 2015
4/29
 * Task:
 * Run 3 trains to see if changing a certain value in the language model has any effect.

5/1
 * Looked at logs and final reports from previous semesters for info on the real time factor

5/4
 * Going to read competition reports to get a better idea of what our report should look like

4/29
 * Results:
 * Ran 3 trains to see if changing a certain value in the language model had any effect. I had a sinus infection and could not make it to this class, but notified my group and Professor Jonas well beforehand.
 * Shared results of the 3 trains with group via the group email chain

5/1
 * There seems to be a lot of inconsistencies with real-time factors even with similar computing conditions

5/2
 * Shared an article about real-time factors posted by IEEE with the group. I was still feeling sick at the time and looking back on it I'm not sure my argument made much sense.

5/4
 * Plan:
 * Reach out to group members working on the competition report to see if me/other members peer-reviewing what we have before we submit it on Wednesday would be helpful
 * Concerns: