×

INDI Library v2.0.6 is Released (02 Feb 2024)

Bi-monthly release with minor bug fixes and improvements

New to Astroberry and weird problem

  • Posts: 5
  • Thank you received: 0
Hello all,

Just started using astroberry (bought a new RPI 3B+ just for that).

Here is my setup:
RPI (astroberry last available edition 2.0.3 + updates) - Connected to the internet via Wifi
USB GPS (working fine, getting Fix and time but for some weird reasons no maps are displayed) Indi driver GPSD
Synscan handheld connected to RPI via USB (control of the AZ-GTI mount working fine via KStars) Indi driver SynScan
ZWO ASI385MC via USB (Now I know the camera is USB3 and RPI 3B+ is USB2 but in all fairness I doubt that this is the issue) Indi driver ZWO CCD as I didn't find any other that would match it, but the control panel recognises the camera as a ASI385MC so I guess it is ok.
Canon 5D Mark III via USB using Indi Canon DLSR driver.

Everything shows up in lsusb -v.

All of this works (seems to work fine - no chance to fully test everything yet).
When in Indi control panel, I can do all the setups for every single bit of kit connected to the RPI and the communications look ok. I can do streaming from the Canon 5D without an issue and even click lights (though I messed up the setup and the lights were not saved at all...)
As soon as I start the streaming for the ASI, I have around 5-10 seconds and the RPI simply crashes, when I say crash I mean a full crash, even the TCP Stack stops working (ssh connection lost, no answers to ping). The only option is a full power cycle of the RPI. And this is systematic, I start the streaming on the ZWO, RPI crash.

Any ideas???

Thanks in advance,
C
3 years 6 months ago #60700

Please Log in or Create an account to join the conversation.

  • Posts: 5
  • Thank you received: 0
So...

I managed to do some further tests (tested the camera on two other devices and it works fine), also swapped various components (raspberry, power supply, SD card...)in the setup just to rule out a hardware issue.

I did however manage to finally get something that may give a clue to the contributors of indilib.

This was in the syslog
Sep 29 10:15:58 astroberry kernel: [  158.534490] usb 1-1.2: reset high-speed USB device number 4 using dwc_otg
Sep 29 10:16:14 astroberry kernel: [  174.544835] usb 1-1.2: reset high-speed USB device number 4 using dwc_otg
Sep 29 10:16:33 astroberry kernel: [  193.234870] usb 1-1.2: reset high-speed USB device number 4 using dwc_otg
Sep 29 10:16:52 astroberry kernel: [  212.234867] usb 1-1.2: reset high-speed USB device number 4 using dwc_otg
Sep 29 10:17:01 astroberry CRON[1469]: (root) CMD (   cd / && run-parts --report /etc/cron.hourly)
Sep 29 10:17:05 astroberry kernel: [  225.792931] dwc_otg: DEVICE:004 : update_urb_state_xfer_comp:751:trimming xfer length
Sep 29 10:17:08 astroberry kernel: [  228.005826] 8<--- cut here ---
Sep 29 10:17:08 astroberry kernel: [  228.005863] Unable to handle kernel paging request at virtual address ffe0ffff
Sep 29 10:17:08 astroberry kernel: [  228.005879] pgd = c3819a2e
Sep 29 10:17:08 astroberry kernel: [  228.005887] [ffe0ffff] *pgd=33fde861, *pte=00000000, *ppte=00000000
Sep 29 10:17:08 astroberry kernel: [  228.005911] Internal error: Oops: 37 [#1] SMP ARM
Sep 29 10:17:08 astroberry kernel: [  228.005920] Modules linked in: cmac rfcomm bnep hci_uart btbcm bluetooth ecdh_generic ecc fuse pps_ldisc pps_core binfmt_misc 8021q garp stp llc spidev brcmfmac cdc_acm brcmutil sha256_generic libsha256 raspberrypi_hwmon cfg80211 rfkill bcm2835_codec(C) i2c_bcm2835 v4l2_mem2mem bcm2835_isp(C) bcm2835_v4l2(C) bcm2835_mmal_vchiq(C) videobuf2_dma_contig videobuf2_vmalloc videobuf2_memops videobuf2_v4l2 videobuf2_common snd_bcm2835(C) snd_pcm spi_bcm2835 videodev snd_timer mc snd vc_sm_cma(C) uio_pdrv_genirq uio fixed i2c_dev ip_tables x_tables ipv6 nf_defrag_ipv6
Sep 29 10:17:08 astroberry kernel: [  228.006027] CPU: 3 PID: 1324 Comm: indi_asi_ccd Tainted: G         C        5.4.51-v7+ #1333
Sep 29 10:17:08 astroberry kernel: [  228.006031] Hardware name: BCM2835
Sep 29 10:17:08 astroberry kernel: [  228.006055] PC is at __kmalloc+0xac/0x31c
Sep 29 10:17:08 astroberry kernel: [  228.006062] LR is at 0x8e963d58
Sep 29 10:17:08 astroberry kernel: [  228.006068] pc : [<802dded4>]    lr : [<8e963d58>]    psr: a0000013
Sep 29 10:17:08 astroberry kernel: [  228.006072] sp : 8e963d58  ip : 8e963d58  fp : 8e963d9c
Sep 29 10:17:08 astroberry kernel: [  228.006076] r10: b2ba99c0  r9 : b3001b00  r8 : 806ae9b8
Sep 29 10:17:08 astroberry kernel: [  228.006082] r7 : 00000cc0  r6 : 00000100  r5 : b3001b00  r4 : ffe0ffff
Sep 29 10:17:08 astroberry kernel: [  228.006087] r3 : 00000000  r2 : 80ca7310  r1 : 32b39000  r0 : 00000000
Sep 29 10:17:08 astroberry kernel: [  228.006093] Flags: NzCv  IRQs on  FIQs on  Mode SVC_32  ISA ARM  Segment user
Sep 29 10:17:08 astroberry kernel: [  228.006098] Control: 10c5383d  Table: 1db1806a  DAC: 00000055
Sep 29 10:17:08 astroberry kernel: [  228.006103] Process indi_asi_ccd (pid: 1324, stack limit = 0x8c31e949)


I have the full stack that comes after that if required for further debugging, but it does look like the indi_asi_ccd did something wrong, I wonder if it is because the ASI385MC is not really handled by it.

Anyone could provide me some help ?

Best regards,
C
3 years 5 months ago #60812

Please Log in or Create an account to join the conversation.

  • Posts: 5
  • Thank you received: 0
So here is a more precise diagnostic (I got a bit of time).

The streaming crashes the RasPi (via KStars or indilib config local or remote) only when the format is set to RAW 16 bits... the other three are working just fine... I have deliberately left the streaming on using RAW 8 bit for the past hour without a problem and did 15 minutes on RGB 24 and Luma.

Whilst that solves (ish) my problem, it'll be interesting to understand why RAW 16 bits preview crashes the RasPi.

So to all ASI users here do you experience crashes when using the RAW 16 bits streaming ? It'll be interesting to see if it is a general ASI problem or linked to specific camera models...

BR,
C
Last edit: 3 years 5 months ago by Ced. Reason: Formatting issues
3 years 5 months ago #60818

Please Log in or Create an account to join the conversation.

  • Posts: 37
  • Thank you received: 5
Hi,

How much memory does your Pi have? I suspect 1 or 2gb, and FITS viewer is quite demanding.

I am using a 4gb RPI4 and had similar issues (though my images are 120mb). I have managed to get FITS viewer running by adding maximum swap and zram.

However I noticed it slowed down my sequence waiting for image to load. So I have completely deactivated it, which I advice on a RPI.

PS: I am using an ASI6200, so yes I use 16bits RAW
Last edit: 3 years 5 months ago by Palmito. Reason: Stroke out useless question
3 years 5 months ago #60820
Attachments:

Please Log in or Create an account to join the conversation.

  • Posts: 1957
  • Thank you received: 420
In his original message he says RPi 3B+ so 1 Gb.
3 years 5 months ago #60822

Please Log in or Create an account to join the conversation.

  • Posts: 37
  • Thank you received: 5

Sorry about that, I thought revision 3B+ had several memory configurations, my mistake....
3 years 5 months ago #60824

Please Log in or Create an account to join the conversation.

  • Posts: 1957
  • Thank you received: 420
No need to apologize :)

Anyway, I think you are right that there isn't enough memory for streaming at that bit depth. At least 4Gb and possibly 8 would be necessary so that means getting an RPi4.
3 years 5 months ago #60827

Please Log in or Create an account to join the conversation.

  • Posts: 5
  • Thank you received: 0
Correct, it is a RasPi 3B+

The fun fact (if that can be called fun), is that if I do captures, everything works fine (or at least so it seems) the images are not that big (the 385 is 1080p so not that big), it is just activating the live view (or streaming) which is quite convenient as the ASI is my main camera and I do not have an auto focuser yet...
And when using my Canon 5D the streaming does work just fine even using maximum resolution. The only difference is - drivers aside of course - that the canon 5d has its own power and the ASI takes the power from the RasPi USB, I have a usb hub with external power on the way, I'll see if that changes the behaviour. But the power indicator on the RasPi does not blink at all so I guess from a power supply perspective it should be good.

So being able to view the picture whilst focusing is what I am looking for really.
I guess just choosing temporarily the Raw 8 bits format and doing a live view whilst focusing and then stopping live view and revert to 16 bits is a workaround...
I just need to remember to change the formats :)

BR,
C
3 years 5 months ago #60828

Please Log in or Create an account to join the conversation.

  • Posts: 37
  • Thank you received: 5
FITS viewer performs quite a lot of things when displaying an image.
The 5D mk4 is 10 bit, so I guess the bit depth is the issue.

I had the same issues you had, but only in the FITSViewer window. But the preview in the focus tab works flawlessly even in 16bit RAW and of course even if the FITS viewer is deactivated in options ;-)
So you should be able to focus from there without changing to 8bit RAW.

I like to monitor my subs while they are acquired. As a workaround I simply open fits files from the file browser.
Beware it will open a second instance of KStars, so you might need to extend your swap an zram partitions ;-)
Last edit: 3 years 5 months ago by Palmito.
3 years 5 months ago #60830

Please Log in or Create an account to join the conversation.

  • Posts: 5
  • Thank you received: 0
Thanks so much everyone...

Just to stir it a bit more... whilst the RasPi is doing the Indi lib, I also tried to run KStars on my main computer (I9 9900K, 64 GB RAM) connecting to indilib on RasPi... and the result is the same... RasPi Crash on Raw 16 Bits Streaming... And in that case there is a lot less load on the RasPi.

But hey, I have a workaround for now :) ...

Looking forward to the next clear night sky...

BR,
C
3 years 5 months ago #60831

Please Log in or Create an account to join the conversation.

  • Posts: 37
  • Thank you received: 5
Yes I would have thought the same. But maybe not.

Considering your pixel count I doubt, but I could be wrong, it will be useful for your case, but with the ASI6200 I had to increase allowed memory for INDI drivers.

On astroberry I edited:
/var/www/indiwebmanager/indi_server.py

and increased memory to 160MB (default was 100):
def __run(self, port):
        cmd = 'indiserver -p %d [b][i]-m 160[/i][/b] -v -f %s > /tmp/indiserver.log 2>&1 &' % \
            (port, self.__fifo)
        logging.info(cmd)
        call(cmd, shell=True)
Last edit: 3 years 5 months ago by Palmito. Reason: formatting gone wrong
3 years 5 months ago #60837

Please Log in or Create an account to join the conversation.

Moderators: Radek Kaczorek
Time to create page: 0.821 seconds