Speech:Spring 2013 Tyler Martin Log


 * Home
 * Semesters
 * Spring 2013
 * Proposal
 * Report

Task

 * Review current hardware specification
 * Get Caesar up and running
 * Research fans to replace broken ones
 * Send options for replacements to Mike
 * Research LDAP
 * Note: Mediawiki uses LDAP

Results
Wednesday 1/30
 * Got Internet Connection to Server Room
 * Caesar
 * Plugged in wires and turned server on.
 * When turned on there was an error for fans 3 and 4
 * Turned off server and took apart.
 * Located fans 3 and 4 and noticed they could not spin.
 * Took fans apart. Group decided we needed new ones.
 * Began research on fans for Caesar.

Friday 2/1
 * Gave SpEAK Experiment group: Scott, Nick and Marc access to network drive.
 * Scott asked for more access and SQL (told him I would get back to him)
 * Continued research on fans for Caesar

Monday 2/4
 * Submitted recommendations for fans to Mike
 * Had short discussion with group about logs
 * SpEAK Experiment Group Acess:
 * Gave Scott remote desktop access and SQL credentials
 * Waiting to hear back if they worked to add other users.

Tuesday 2/5
 * Read through logs today
 * Reviewed hardware configurations for Caesar

Plan

 * To look into LDAP for next week as did not have the time this week
 * Add remaining users to STEM so they can start their work
 * Install fans in Caesar if they come in
 * Come up with plans for hardware upgrades (if needed)

Concerns

 * LDAP make time some time or may be hard to do, but up for the challenge

Task

 * Caesar fans
 * Add replacement fans into Caesar
 * Make sure fans are working properly
 * Make adjustments if fans are not working properly
 * Come up with plan with group for what we plan to accomplish
 * Look into hard drives given to me by Mike
 * Help Mike Bi. with version difference options
 * Help Brian look into backup solutions
 * Review more OpenSUSE LDAP configuration guides

Results
Saturday 2/9
 * Continued reviewing hardware configurations for Caesar and other servers
 * Looked a guides for setting up LDAP on OpenSUSE (OS Caesar runs).
 * Started looking at drives given to me by Mike.
 * No progress to report here yet.
 * Older drives (IDE) looking to see if have parts at home to make machine
 * Sent group a message we need to have a discussion to set down a plan for our goals.
 * Reviewed what Mike Bi. currently has for differences in versions for OpenSUSE

Sunday 2/10
 * Reviewed logs today.
 * Emailed group to check progress of items for week
 * Saw Brian's review on CloneZilla
 * Offered some information from personal use
 * Compression is good
 * Emailed Scott with update on STEM issues

Monday 2/11
 * Caesar fans
 * Attempted to add fans into Caesar today
 * Housing prevented new fans to fit properly. (We cut housing)
 * Fans when plugged in did not turn on.
 * Tested with other fans working inside Caesar (in case it was the board port) Other fans worked.
 * Used spare wire to change connections to board by adding into connection on fan (got a working combination, fans need resoddering)
 * Reconnected Caesar
 * Discussed Clonezilla some more with Brian
 * Made sure to let group know to keep up with logs
 * Started working on hard drives to see if they are salvageable
 * Report on these to come

Tuesday 2/12
 * More discussion with Brian about Clonezilla and how it works
 * Reviewed proposal Brian drafted up for group
 * Received email Caesar was not accessible
 * No response from Caesar when tried to wake up
 * Restarted Caesar to receive message that there are bad sectors on one of the drives
 * Turned off and reseated hard drives
 * Turned back on still no response
 * Status screen still reports two fan fail and now drive fail
 * Read over more LDAP
 * Looked at failed drives from Mike
 * One clicks, one cannot format and one does not boot
 * Research founds that boards on drives can fix some issues with drives
 * Reviewed others logs for the week

Plan

 * Reinstall fans into Caesar
 * Investigate drive failure on Caesar
 * Rebuild system if total drive fail
 * Propose some backup and attempt to implement it
 * OS upgrades if have to rebuild system

Concerns

 * Data may be lost with the drive failure in Caesar.
 * Active Directory may not be able to be configured 100% properly to suggested way

Task

 * Get Caesar Working
 * Reinstall OS
 * Ensure filesystem still intact
 * Get network configured properly
 * See if LDAP Is potentially even possible for Caesar and STEM SpEAK page
 * Work with Brian on proposal for Clonezilla

Results
Wednesday 2/13
 * Caesar
 * Ran error checks with group and Eric and Matt on drives
 * No problems were reported with 36GB drive that had OS
 * Tried rescue disk on the drive to see if we could see filesystem
 * Could not mount drive
 * All files were on RAID array on other drive only OS on corrupted drive
 * Installed new 73GB SHCI drive.
 * Installed 12.2 OpenSUSE on Caesar
 * Had issues with some of setup features
 * Could not get network to work properly
 * Downgraded to 11.3
 * Network works and can SSH in
 * Need to fix from DHCP to Static IP
 * Wrote back to Vinnie back project proposal
 * Discussed alittle more Clonezilla with Brian

Sunday 2/17
 * Read through few logs that have been posted
 * Noticed that Mike's plan for temporary alternatives systems was approved
 * Going to email Brian to see how proposal for Clonezilla backup is going.
 * Researched static IP issues for Linux
 * Will need to look specifically for OpenSUSE here.
 * Will email Eric and Mike to see what is left to be configured on Caesar to restore proper working order

Monday 2/18
 * Caesar Configuration
 * Worked with Eric again today to try and fix some issues with Caesar
 * Static IP address issue was resolved. DHCP has a set static on its end but prefer one on ours
 * Default Gateway was added to network settings for Eth0 under YaST
 * Different nameserver was added.
 * UNH has several different name servers.
 * Cannot remember exactly which is which but I believe it was changed to the Durham one which works better.
 * Google added as a nameserver as well for test purposes
 * Caesar was added to the WILDCAT (Windows) Domain.
 * Users could login using Blackboard credentials still needs to be configured more
 * This needs to be documented in information section still.
 * Talked to Matt about NTP with Caesar and how it was not set properly
 * Started downloading some other Linux distributions for testing purposes

Tuesday 2/19
 * Read through some logs today
 * Installed updates on Caesar
 * Looked into more OpenSUSE active directory on Caesar
 * Added NTP to startup services
 * Read through Brian's proposal for Backups

Plan

 * Try and implement alternative systems to see if we can find a better OS than OpenSUSE
 * Test out backup solution on a system with RAID array
 * Continue configuring Active Directory to specifications on Caesar
 * Finish analysis on potentially bad hard drive from Mike

Concerns

 * Our proposed backup solution may not work well with RAID arrays

Task

 * Look at further implementation guides on active directory for Linux
 * Test implementation on OpenSuse to see if can use for authentication
 * Try clonezilla backup on another system that has Raid drives to see if it works

Results
 Saturday 2/23
 * Read logs
 * Followed up with group members about emails sent out

 Sunday 2/24   nothing to report on integration yet that hasn't already been discovered 
 * Read proposal Mike made for alternative OSes
 * Looked into further integration of active directory for OpenSUSE
 * Looked into ways to integrate in proposed systems by Mike.

 Monday 2/25 
 * Added Caesar back to Windows Domain
 * Tested regular login after adding to domain to ensure SSH access still works
 * Active Directory Login should work with reboot of system.
 * Will create directories in on local OS drive in a WILDCAT or AD folder depending on credentials
 * Looking at mapping this to the directory on the mount drive
 * Looked at Fedora 18 (new release) for Active Directory support
 * Has support out of box
 * Created virtual machine on personal computer with Fedora 18 to test environment and how active directory works tomorrow

 Tuesday 2/26 
 * Downloaded Fedora 18 (several versions to install virtual machine)
 * Was not able to install in a virtual machine to test active directory
 * Installer would fail when choosing settings before copying over files.
 * Note sure what this will mean for a test main with Caesar's hardware (tried installing on newer machine)
 * Read logs to see what others had done for the week.
 * Checked to make sure Caesar was still online no hickups in service

Plan

 * Test out Clonezilla with system downtime for when doing other test installs
 * Swap out all drives in Caesar and try out some proposed test OS's that Mike looked into
 * Record outcomes of each test OS
 * Try Active directory login on each and see if easier solution then OpenSUSE

Concerns

 * Fedora 18 may not be a viable option for Caesar with the problems had with installing in virtual machine
 * May not be able to change the folder made by Linux for AD login as it creates a WILDCAT folder for user folders

Task

 * Test Fedora in a virtual machine
 * Document current active directory progress in Information section of wiki

Results
Friday 3/1
 * Created virtual machine on home machine
 * Installed basic GNOME based Fedora 18
 * Got it working this time
 * Installed OS updates
 * Explored options of configuring options we use on OpenSUSE
 * Tried to join active directory
 * Did not want to work or was taking too long with VPN
 * Has builtin support right out of the box

Saturday 3/2
 * Read through logs today
 * Let my team know where I was at so far
 * Asked team where they were and if needed help

Sunday 3/3
 * Created page for Active Directory
 * Made appropriate links following structure of how Prof Jonas set it up
 * Link had to be added to each page of project notes to have proper structure
 * Had to be added on information page
 * Side note added to include my name will include Eric's as well as he was part of setup
 * Began adding information of active directory discovery to information page
 * Will list problems that we may face for further implementation
 * Solutions to problems if found will also be added
 * Let Mike Bi know which version of PERL is running on system (command is "perl -v")

Tuesday 3/5
 * Read logs
 * Emailed Mike Bi to see if he sent in proposal for test system
 * Created Fedora 18 disc for potential test system tomorrow

Plan

 * If proposal approved work out a day (Wednesday potentially day of class to try Fedora 18)

Concerns
Fedora test environment may not work well with equipment like OpenSUSE 12.2

Task

 * Test Fedora 18 on Caesar with other HDDs
 * Install new memory with Mike when it comes in
 * Results to be posted and information page to be changed
 * Finish active directory page by next class meeting
 * Create Clonezilla boot disk

Results
Wednesday 3/6
 * Showed group how to boot clonezilla
 * Told Brian to try his own test environment
 * We then can document the process for future classes
 * Removed all of Caesars drive and added different ones to test Fedora 18
 * One disc in new test set failed created two containers similar to Caesar
 * Installed Fedora 18 XFCE without a problem
 * Graphics issue until you install updates
 * Easy to set static IP and SSH outside firewall
 * Let Eric know Caesar would be done tomorrow for memory maintenance so he did not start a train

Thursday 3/7
 * Memory upgrades
 * Powered off each machine
 * Took it out of rack and opened up
 * Did memory configurations to all so that batch machines now all have 4GB
 * Note: OS only recognizes 3.9GB of it as they are 32-bit machines
 * Attempted to add memory to Caesar from memory from another server in server room
 * Memory seemed to be no good
 * Caesar would not boot so new memory removed
 * KVM switches removed from rack.
 * Belkin KVM switch from extra rack added to Caesar
 * All machines had to be rewired looks cleaner in back

Monday 3/11
 * Read logs

Tuesday 3/12
 * Read logs

Plan
Wrap up systems group items as we will be divided up next meeting to begin a modeling a group

Concerns
A backup should be made of Caesars OS drive so that incase of another failure it can be quickly restored to its current state

Task

 * Read new guide from modeling group about running a train
 * Meet with new group to learn how to run a train
 * Document any errors along the way

Results
Friday 3/22
 * Added labels on new KVM switch in Capstone rack to identify which machine goes with each button.
 * Switch may need to be flashed if want the screen to show name of system as well like old switches did.
 * Began reading up on modeling process and reviewing information sections.
 * Started to brush back up on UNIX commands as they will be needed when learn how to do a train from Tom on Sunday.

Sunday 3/24
 * Read through Modeling Groups changes to the Run A Train Page.
 * New page is very straight forward and has error handling as you go.
 * There is some errors in the code itself minor and will be noted to them.
 * Had hangout with Tom, Scott and Vincent
 * Each created an experiment directory to test a train run
 * I created experiment 31 in the filesystem.
 * Following guide and Tommy's instruction was able to setup the experiment
 * We made minor changes to the guide as there was a few small errors
 * Was able to run my train took around 10 minutes for a tiny train.
 * Tom and Scott were not able to run theirs because of problems with their systems we believe
 * Created Experiment 0031 page for documentation

Monday 3/25
 * Received email that Brian had put up instructions for how to use Clonezilla.
 * Looked through his instructions pretty straightforward similar to what I have done.
 * Copied his instructions over to Backup page.
 * Restructured the Backup page
 * Created a small history section at the beginning of page of known backup history.
 * Moved Bethany's Google Code writeup to a headered section called Google Code.
 * Created section for Clonezilla adding in Brian's steps.
 * Made changes to steps and formatting of text.
 * Let systems group know of the changes made.
 * Read how to create a language model.

Tuesday 3/26
 * Read logs
 * Read how to run a decode.

Plan

 * In class go over language model and decode with group
 * Fill in others that missed hangout about what to do

Concerns

 * That all batch machines are not 100% configured the same as we did run into some errors on some of the machines when we had our Google hangout and tried to run a train.

Task

 * Hangout with group on Google on Sunday to review status and continue with language model and decode
 * Catch up group members that are behind
 * Hopefully get a full decode done successfully.

Results
Sunday 3/31 UPDATE 3/31
 * Google Group Hangout
 * Went over where everyone was in modeling
 * Some are farther behind then others.
 * Followed guide and was able to do a language model and setup the decode.
 * Decode failed on 0031 as it was a tiny train and we discussed issues last week with it
 * Created new experiment 0063.
 * This train experiment is using the updated guide with a mini.
 * Train ran and completed without stopping for errors.
 * Created language model for this experiment.
 * Decode is still running at the time of this posting.
 * Decode finished with no errors.
 * When trying to score had errors.
 * Errors documented in experiment page.
 * Created Experiment 0063 page for documentation.

Monday 4/1
 * read logs today

Tuesday 4/2
 * Worked with Matt today to try and catch him up to the rest of the group
 * Found out he was looking in the wrong area for the guide.
 * Was able to get him to create an experiment and started him on his way with running a train.
 * Noted to him to use the dictionary from experiement 0058 with Eric's changes before trying to run it
 * Talked with Matt about his transcripts and how he wants to make a new mini with his stuff
 * Read logs today as well

Plan

 * N/A

Concerns

 * That everyone in the class is in a different place.

Task

 * Finish catching up group members on how to train, decode and score.
 * Work with new group to get transcript files for new models.
 * Try to train, decode and score the new five hour.

Results
Wednesday 4/3
 * Work with group to catch up members in group that have fallen behind.
 * Tommy helped Bego, I helped Matt.
 * Created experiment 0065 with Matt.
 * After trying to walk him through the guide, something went wrong and the experiment got messed up.
 * Created experiment 0073 to walk him through the process as he made transcripts to work with next week.
 * This experiment worked very well. Using the dictionary from experiment 0058 we were able to shave some of the time off creation.
 * Note: we had run the dictionary creation in 0065 so Matt understands how it is done.
 * Decode and Scoring were done with this experiment as well.
 * Page for experiment will be created another day.
 * New groups were created.

Sunday 4/7
 * Read logs today
 * Talked with Tommy to see if he heard anything about new train
 * Also discussed if having a group meeting online this week

Monday 4/8
 * Read logs today

Tuesday 4/9
 * Found out that first_5hr was on server
 * Created experiment 0078
 * Everything copied over fine
 * Used new dictionary script that other group created
 * Works much better and gives a text file of missing words unlike old script
 * Was missing 111 Words
 * Worked on finding missing words for dictionary.
 * Found all words few typos need to fix for train to run

Plan

 * Fix missing or mistyped words in dictionary
 * Get train to run
 * Create language model for 5 hour train
 * Decode and score the new experiment

Concerns

 * It will bug out after a few hours wasting time
 * Other group members are behind.

Task

 * Work with group to fix errors in the transcript files
 * Rerun experiment as a group to try and reduce error rate.
 * Get more people involved in weekly discussion

Results
Wednesday 4/10
 * Added missing people to group email so they were not left out of group conversation
 * Finished decoding my experiment 0078
 * Scored experiment without any errors at all.
 * Created experiment pages for 0073 and 0078.
 * 0073 was experiment from last week used to catch Matt up in process of training and decoding
 * 0078 is a 5 hour train that I did myself.

Sunday 4/14
 * Read logs today

Monday 4/15
 * Hangout with group
 * Created experiment 0084
 * Tried new genTrans3.pl script for transcript creation.
 * Script still has bugs as it does not remove everything it should.
 * Adds alot of unnecessary work for dictionary creation.
 * Edited script a few times with Matt to try and get bugs out
 * Missing 132 words.
 * Called it a day

Tuesday 4/16
 * Recreated experiment as 0084 had errors
 * Created 0086 in place of 0084 so error could be looked at later
 * No problems this time creating train
 * Decode took a long time almost 6 hours.
 * Score worked without issue
 * Error result is higher than other experiment of 46%.
 * Created experiment pages.

Plan

 * Figure out why we keep getting such a high error rate

Concerns

 * That genTrans3.pl did not help reduce error rate. Rate increased.

Task

 * Continue working on improving error rate for first five hour train

Results
Saturday 4/20
 * Read logs today

Sunday 4/21
 * Compiled list of missing words from Matt's exp 0085 for group to work on.
 * Shared list in a Google drive document.
 * Some group members had questions about missing words and how to go about finding them.
 * Filled in missing from what I had in other experiments
 * Created Experiment 0088 to try and run train with genTrans5.pl and new dictionary.
 * IN PROGRESS

Monday 4/22
 * Verified that nothing was lost on Caesar with outage of systems
 * Everything seemed to still be intact
 * Powered up systems when the room cooled.
 * Noticed error in creation of language models on my 5 hour trains
 * Used mini language model instead of 5 hour
 * Changed experiment 0088 to use 5 hour LM
 * Reran decode on 0088.
 * Ran train and decode on 0085.

Tuesday 4/23
 * Scored both experiments 0085 and 0088.
 * Got higher error rate 47%
 * Both used genTrans5.pl script for generation
 * Both had a different dictionary from one another
 * Both used 5 hour LM
 * Read logs

Task

 * Divide up the work between group members for final report
 * Work on section of report for draft due Wednesday

Results
Friday 4/26
 * Read logs today

Sunday 4/28
 * Read logs today

Monday 4/29
 * Worked on Google Drive Doc for proposal
 * Added in sections about new scripts created and improvements they made
 * Talked about new corpuses made
 * Wrote about five hour train process.

Tuesday 4/30
 * Read parts of report that others in group wrote up.
 * Made changes where saw fit.
 * Added reports sections and write ups to wiki.
 * For entire class to see current status.
 * Link added to report page to current working Google Doc file

Plan

 * Continue working on final report

Concerns

 * None. Report seems to be going smoothly

Task

 * Continue final report

Results
Saturday 5/4
 * Received email from Eric that Caesar was down again.
 * Found Caesar and the drones offline when got to Pandora.
 * A/C Unit did not fail this time
 * Power Strip tripped itself. Too much power being drawn on the one strip
 * Borrowed a power strip from a CT classroom to split up the power draw
 * Booted the systems back online.
 * Made sure file system was good
 * Notified Eric and Mike that system was good to go.

Sunday 5/5
 * Read logs today

Monday 5/6
 * Continued to work on final report
 * Began reviewing logs more indepth for grading purposes.

Tuesday 5/7
 * Looked over final report document
 * Took changes and added to mediawiki
 * Read over all logs again today for grading tomorrow

Plan

 * Finalize report for next Wednesday

Concerns

 * N/A