Finder Documentation

Äîêóìåíò âçÿò èç êýøà ïîèñêîâîé ìàøèíû. Àäðåñ îðèãèíàëüíîãî äîêóìåíòà : http://www.mrao.cam.ac.uk/~jsy1001/dimmwit/finder/troubleshoot.html
Äàòà èçìåíåíèÿ: Thu Aug 28 13:31:29 2008
Äàòà èíäåêñèðîâàíèÿ: Tue Oct 2 11:53:09 2012
Êîäèðîâêà:
Ïîèñêîâûå ñëîâà: çàïðåùåííûå ñïåêòðàëüíûå ëèíèè

Troubleshooting

Starting Finder

I selected the "All Finders" menu option on an X display, but one of them didn't appear.

Connectivity Issues

When I tried to connect to the telescope controller I saw an error message, " TCP/IP socket error: computer.in.domain.cam.ac.uk: Connection refused."
When I tried to connect to the CCD I saw an error message, " Could not connect to Starlight Xpress CCD ".
I'm using finder across a slow/busy network and the refresh rate is very low.

Crashes

Finder crashed! Waaaaah!
Finder crashed and now I can't interact with anything else on my X display!

Display Strangeness

Finder does not complain when I connect to the CCD, but all I get is a black screen in the video display.
There are strange vertical stripes with irregular spacing on the images and in my files [deprecated].
I tried using 1D DIMM mode (Differential Image Motion Monitor), but my file was ruined by vertical stripes [deprecated].
The video display is showing psychedelic stripe patterns!
The video display is still showing psychedelic stripe patterns!
I just went from DIMM mode to normal fullscreen readout and the display froze.

Unexpected behaviour

Finder was doing subframe readout/tracking when it suddenly stopped and went back to full screen readout all by itself!
I draw a rectangle around the star I want but it disappears as soon as I let go of the left mouse button.
I want to use the [blah] function but it's greyed out.
I can see strange artefacts and they don't move when the telescope moves.

Error logging

Where is the error log?
When I look in .xsession-errors, I see a message, "Xlib: extension "RENDER" missing on display "localhost:10.0"."
How about "Qt: Locales not supported on X server"?

Starting Finder

I selected the "All Finders" menu option on an X display, but one of them didn't appear.

This problem is yet to be resolved. It appears to be related to the local X server having trouble launching many remote applications at once. For now, figure out which telescope finder didn't run and choose the menu item that launches that telescope finder.

Connectivity Issues

When I tried to connect to the telescope controller I saw an error message, "TCP/IP socket error: computer.in.domain.cam.ac.uk: Connection refused."

Most likely this is because the telescope controller doesn't know how to talk to finder, possibly because it isn't running, you are running telescope2/ctl2, or have started telescope3/ctl3 manually. Stop telescope and ctl using the "stop" script on the telescope controller and restart using the "start" script. This will set up the mediating program on the telescope controller computer.

This failure mode is not catastrophic, only features that require communication with the telescope will be disabled. It is still possible to view the star field and acquire manually.

When I tried to connect to the CCD I saw an error message, " Could not connect to Starlight Xpress CCD".

This means that the single board computer running finder and the microcontroller that drives the CCD head are not communicating over the USB link between them. If it is not a physical break, either the microcontroller or the CCD driver modules on the single board computer have crashed. This can happen if there has been a power interruption.

This can be cured by rebooting the single board computer. Follow the cures in the crash section.

I'm using finder across a slow/busy network and the refresh rate is very low.

This is due to limited ethernet bandwidth between you and Lords Bridge. You can improve matters by using the compression option in ssh:

Figure out the name of the computer you want to connect to from this table .
From an xterm, type "ssh -C coast@computername finder ".
The "-C" requests that the ssh server compress the data.

Crashes

Finder crashed! Waaaaah!

At COAST, a finder in a given telescope will sometimes crash when the beam splitter in that telescope is moved in or out, for reasons which are unknown but obviously power related. The problem is particularly acute with the outer west telescope. The symptom is that the finder video and the user interface freezes. Whether you are using a COAST finder or not, rebooting the single board computer almost always works:

Make sure there is no power to the camera head. On COAST telescopes you can do this by moving the beam splitter out.
Figure out the computer name from the title bar of the moribund finder, for example cstfinder4.
From an xterm, type "ssh root@computername reboot " and give the root password (ignore the xauth error message). The root password is as for the other COAST machines.
Wait about half a minute for the computer to boot before trying to run finder again.

If that doesn't work, the following procedure always works:

Make sure there is no power to the camera head. On COAST telescopes you can do this by moving the beam splitter out.
Go out to the telescope and disconnect the power lead feeding the single board computer power socket:

The single board computer is the top board here.

Location of power socket

The power socket on the computer board is here.

At the other end of the telescope, disconnect all leads at the USB interface:

The power lead to the transformer.
The lead to the camera head.
The USB lead.

Wait a few seconds so that the USB interface forgets its neurotic little program.
Reconnect all the leads to the USB interface:

The USB lead.
The lead to the camera head.
The power lead to the transformer (which should be off because the beam splitter is out).

Reconnect power to the single board computer.

You should see some pretty flashing lights.

Wait about half a minute for the computer to boot before trying to run finder again.

Finder crashed in 1D DIMM mode and now I can't interact with anything else on my X display!

This problem occurs when a pull-down menu is opened in finder while it is displaying in 1D fast readout mode and this menu obscures the fast readout display. On rare occasions, the display freezes and the instance of finder keeps the focus, which means that the user cannot interact with any other displayed applications on their X console.

It turns out that finder is still functioning perfectly well but for some reason (network problem?) the Xserver/client dialogue has become confused. You can work around the problem as follows:

The internet name of the offending finder is displayed in its title bar, for example apfinder4.
From an xterm elsewhere, log into the offending finder as coast.
Type "ps" to find out the finder process ID.
type "kill -s SIGTSTP <finder processID>

Finder should then be functional again and you will have regained control of your X desktop. I don't understand the problem or why this works, but it does!

To prevent the problem from recurring, in 1D fast readout mode you should use the buttons in the toolbar and not the pulldown menus.

Display Strangeness

Finder does not complain when I connect to the CCD, but all I get is a black screen in the video display.

The camera head has no power or is not plugged in. If you are using a COAST telescope, try moving the finder beamsplitter in using the telescope controller.

There are strange vertical stripes with irregular spacing on the images and in my files [deprecated].

Subframe image of brick wall illustrating the artefact.

This used to happen with some HX516 cameras. It was eventually discovered that the CCD horizontal clock signal needed for the HX516 is the inverse of that for the MX516 (for which the software was originally written). This is an undocumented "feature". Inverting the signal has cured the problem.

I tried using 1D DIMM mode (Differential Image Motion Monitor), but my file was ruined by vertical stripes [deprecated].

DIMM artefacts.

This problem, which only manifests with HX516 cameras, has the same cause and cure as the vertical stripes problem above.

The video display is showing psychedelic stripe patterns!

The microcontroller-to-single board computer dialogue has become confused. This has been observed with the outer west telescope and happens there when the beam splitter is moved out, cutting power to the camera head. The cause is unknown.

This problem can be cured reliably by rebooting the troublesome single board computer.

Using the controls in the finder GUI:

Disconnect from the CCD.
Disconnect from the telescope if connected.
Quit finder.

Now follow the instructions in the crash section.

A workaround that prevents this problem from happening in the first place is to leave the camera head powered on all the time.

The video display is still showing psychedelic stripe patterns!

Try extinguishing your spliff, you're not supposed to smoke in the bunker anyway :)

I just went from DIMM mode to normal fullscreen readout and the display froze.

The symptom here is that an old fullscreen picture (the last one before DIMM mode started) is displayed, with some banding along the top line or two containing DIMM data. The display doesn't update.

It is likely that the microcontroller is still in DIMM mode, even though finder has told it to go back to normal readout. On the rare instances when this has been observed at COAST, it was cured by going back into the DIMM mode that was used previously and then trying normal fullscreen readout again. If that doesn't work for you, ensure that you have disconnected from the CCD and quit finder. Then reboot the single board computer by following the instructions in the crash section.

Unexpected Behaviour

Finder was doing subframe readout/tracking when it suddenly stopped and went back to full screen readout all by itself!

This is intentional behaviour. Finder will revert to full frame readout without tracking if it detects that the telescope secondary or beam splitter have moved. This stops finder from sending bad offsets to the telescope during such optical changes. As the user will almost certainly want a full frame display after such an event, finder does that too.

I draw a rectangle around the star I want but it disappears as soon as I let go of the left mouse button.

The beamsplitter is out or moving, and/or the secondary is moving. Finder will not let you track or define a tracking region in these circumstances.

I want to use the [blah] function but it's greyed out.

Either Finder is not letting you do that right now (it won't let you track unless you've connected to the CCD first, for example) or your function has not yet been implemented and is disabled to thwart compulsive knob twiddlers.

I can see strange artefacts and they don't move when the telescope moves.

You are running finder with a long exposure? Then there may be trouble with stray light leaking into finder from the bunker (yes, it's that sensitive!). Try turning the bunker lights out. However, stray laser light from the metrology has been seen with finder and more strenuous measures might have to be taken to get rid of that. Some trolley positions are more troublesome than others so if you can, try changing the overall trolley offset.

Error Logging

Where is the error log?

If you have started finder from the menu, all stderr messages end up in .xsession-errors.

When I look in .xsession-errors, I see a message, "Xlib: extension "RENDER" missing on display "localhost:10.0"."

This message, which may be generated when using an older X display, can be safely ignored. Finder expects its X server to include the "RENDER" extension, which is used for things like antialiased text. If finder does not find the extension, it will render its display without it.

How about "Qt: Locales not supported on X server"?

You can also safely ignore this message. It just means that the Qt graphics library that finder uses doesn't know where in the world it is running, so it doesn't know which language to use. As finder only contains English text anyway, this makes no difference to its behaviour.

Last updated 20th October 2004 by bodie