CHAMP bugs to fix: Difference between revisions

From Monnier Group Research Wiki
Jump to navigationJump to search
← Older edit
No edit summary
Monnier (talk | contribs)
0 edits
No edit summary
 
(5 intermediate revisions by one other user not shown)
Line 1: Line 1:
== General ==
== General ==


* ssh error "Disconnecting: Corrupted MAC on input.". Python guis will crash with this error from time to time if they are on ssh (i.e. not user is ssh-ing to wolverine). FB tried debugging using ssh -vvv but was unsucessful. Seems to be hardware+kernel related and require changes with ethtool, see https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.17/+bug/60764 . Currently testing if this solution ( ethtool -K eth0 rx off tx off ) works. If yes, then we can add this as a script to /etc/network/if-up.d/. I suggest we test the md5 sum of transfered large files to test the reliability of the network (can be generated by dd if=/dev/urandom of=bigfile bs=1G count=60).
* ssh error "Disconnecting: Corrupted MAC on input.". Python guis will crash with this error from time to time if they are on ssh (i.e. not user is ssh-ing to wolverine). Annoying as MIRC+CHAMP needs ssh-ing. FB tried debugging using ssh -vvv but was unsucessful. Seems to be hardware+kernel related and require changes with ethtool, see https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.17/+bug/60764 . Currently testing if this solution ( ethtool -K eth0 rx off tx off ) works. If yes, then we can add this as a script to /etc/network/if-up.d/. I suggest we test the md5 sum of transfered large files to test the reliability of the network (can be generated by dd if=/dev/urandom of=bigfile bs=1G count=60).
One potential issue is that transfer rate is supposed to get slower on some NIC, though my tests on wolverine <-> mirkwood scp exchanges show this is not the case.


== CHAMP server ==
== CHAMP server ==


* Critical bug: the ftnums become negative after a while, as shown on the FT gui. -- COULD NOT REPLICATE
* Critical bug: the ftnums become negative after a while, as shown on the FT gui. -- FIXED ?


== Spooler ==
== Spooler ==


* Missing uv and star info -- FIXED
* HA not displayed correctly when negative ?


== FT GUI ==
== FT GUI ==


* Sometimes AMPvsOPD freezes the display completely. Seen during July 2011 run. - COULD NOT REPLICATE
* Hardcoded values: numberOfFrames


== Align GUI ==
== Align GUI ==


* Second display is not opened at ALIGN START -- FIXED
* FPA tab does not seem to do anything anymore
 
* Second display freezes at the end of the alignment sequence. -- FIXED
 
* Need shutter control toggle -- FIXED


* Need better way to move between alignment steps - ONGOING
* Need better way to move between alignment steps - ONGOING
Line 32: Line 26:


* Need useful quadrant mode in wide field view - ONGOING
* Need useful quadrant mode in wide field view - ONGOING


== Other GUIs ==
== Other GUIs ==
== Random List from 2013 Run ==
* default values of shuttermatrix
* get telescope beram order from chara shomehow for mirc and champ, including new buttons on guis.
* proper slts/prsm labels in fiberexplorer
* fix defaults  (ftgui, mric shutter matrix)
* mirc: save xy and Z of fiber positions
* save POPS in mirc header
* mirkwood server should start w/0 vme on
* mirkwood spooler should run w/o all servers up (opler/tiptilt, etc)
* (recent) Delay line positions
* '''FIXED''' fix zeros in uv coordinates FIXED!!
* send ople a target time instead of 0 0 0 0 0 0 (not needed.. just use the HA which is returned)....
* mirc and champ: replace pulldown menus and text boxes with other elements.
* champ: ability to algin in wide field mode.
* champ. fix stop exposure crap (hard)
* champ. save syncronized spooler data even for large strokes
* champ. test phase tracking mode
*champ filter SPIKES out of amplitude(t) gui plots
* make sure search to semi-lock can catch spikes of good seeing (lucky fringes) but
not get screwed up by huge spikes caused by electronics. currently it seems the system misses real fringes probably due to
spike suppression algorithm being too primitive.
* improve usability of champ gui.  the top part of the gui should not be so sluggish. only the plots need to have a timer, no the other widgets.
* fringe 6 B5/B6 pull down state stays in search , doesn't go to semi-lock visually
* whens aving data, the amp[opd] of first BC is corrupted but seems to not affect tracking.
* change colors of AMP(OPD) from red/blue to something that doesn't appear match the flux(OPD) graphs
*champ. add more readout modes.. shorter than 30micron stroke, slower than 250 hz. we need a 'phase trakcing' mode this probably like 20 microns.

Latest revision as of 01:54, 20 July 2013

General[edit]

  • ssh error "Disconnecting: Corrupted MAC on input.". Python guis will crash with this error from time to time if they are on ssh (i.e. not user is ssh-ing to wolverine). Annoying as MIRC+CHAMP needs ssh-ing. FB tried debugging using ssh -vvv but was unsucessful. Seems to be hardware+kernel related and require changes with ethtool, see https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.17/+bug/60764 . Currently testing if this solution ( ethtool -K eth0 rx off tx off ) works. If yes, then we can add this as a script to /etc/network/if-up.d/. I suggest we test the md5 sum of transfered large files to test the reliability of the network (can be generated by dd if=/dev/urandom of=bigfile bs=1G count=60).

One potential issue is that transfer rate is supposed to get slower on some NIC, though my tests on wolverine <-> mirkwood scp exchanges show this is not the case.

CHAMP server[edit]

  • Critical bug: the ftnums become negative after a while, as shown on the FT gui. -- FIXED ?

Spooler[edit]

FT GUI[edit]

  • Hardcoded values: numberOfFrames

Align GUI[edit]

  • FPA tab does not seem to do anything anymore
  • Need better way to move between alignment steps - ONGOING
  • Need calibration procedure - ONGOING
  • Need spiral search - ONGOING
  • Need useful quadrant mode in wide field view - ONGOING


Other GUIs[edit]

Random List from 2013 Run[edit]

  • default values of shuttermatrix
  • get telescope beram order from chara shomehow for mirc and champ, including new buttons on guis.
  • proper slts/prsm labels in fiberexplorer
  • fix defaults (ftgui, mric shutter matrix)
  • mirc: save xy and Z of fiber positions
  • save POPS in mirc header
  • mirkwood server should start w/0 vme on
  • mirkwood spooler should run w/o all servers up (opler/tiptilt, etc)
  • (recent) Delay line positions
  • FIXED fix zeros in uv coordinates FIXED!!
  • send ople a target time instead of 0 0 0 0 0 0 (not needed.. just use the HA which is returned)....
  • mirc and champ: replace pulldown menus and text boxes with other elements.
  • champ: ability to algin in wide field mode.
  • champ. fix stop exposure crap (hard)
  • champ. save syncronized spooler data even for large strokes
  • champ. test phase tracking mode
  • champ filter SPIKES out of amplitude(t) gui plots
  • make sure search to semi-lock can catch spikes of good seeing (lucky fringes) but

not get screwed up by huge spikes caused by electronics. currently it seems the system misses real fringes probably due to spike suppression algorithm being too primitive.

  • improve usability of champ gui. the top part of the gui should not be so sluggish. only the plots need to have a timer, no the other widgets.
  • fringe 6 B5/B6 pull down state stays in search , doesn't go to semi-lock visually
  • whens aving data, the amp[opd] of first BC is corrupted but seems to not affect tracking.
  • change colors of AMP(OPD) from red/blue to something that doesn't appear match the flux(OPD) graphs
  • champ. add more readout modes.. shorter than 30micron stroke, slower than 250 hz. we need a 'phase trakcing' mode this probably like 20 microns.
Retrieved from "https://monnier.astro.lsa.umich.edu/research/index.php?title=CHAMP_bugs_to_fix&oldid=2719"