SLUG stops working

[bmoffitt]

I bought a SLUG and WH-1080 weather station from Ambient weather several months ago. It worked well for several months, and then I lost the signal to the console. It turned out to be a bad xmitter on the weather station, which was replaced by the folks who make it.

However, it took me a month or so to debug the problem (I wasn't exactly putting my full energies into it). By the time I got it working again, my SLUG wasn't working any more.

I have tried re-flashing the OpenSLUG software (2.3 from the archives right here at meteohub.de) and re-building the 2 GB USB flash drive (using the 4.5 firmware). The problem remains.

I have the station set up, along with the sensors. Setting up the sensors, they indicated they were getting good data, and there is data there when I first reboot the system. However, a short time later, when I look at Sensor Setup, I see this message:

error connect socket 127.0.0.1:5555 :Connection refused
Warning: Data logging has been halted.

Any time I go to "Inspect Data" there is no data.

There are no entries in the alarm log, and nothing apparently pertinent in the system log. In the meteohub log, however, there were entries about input/output errors when reading or writing to files in /data/weather. A sample from the log:

logger (07.11.2009 03:07:21): connect station 0 (WH-1080 via USB HID).
wmr928eval (07.11.2009 03:08:01): error while opening file /data/weather//seq-min1 for writing: Input/output error: Input/output error
wmr928eval (07.11.2009 03:09:01): error while opening file /data/weather//seq-min1 for writing: Input/output error: Input/output error
wmr928eval (07.11.2009 03:10:01): error while opening file /data/weather//seq-min1 for writing: Input/output error: Input/output error
wmr928eval (07.11.2009 03:10:02): error while opening file /data/weather//hist-day1 for writing: Input/output error: Input/output error
wmr928fulleval(07.11.2009 03:10:02): recomputation of weather data started.
wmr928eval (07.11.2009 03:10:02): error while opening file /data/weather//hist-hour1 for writing: Input/output error: Input/output error
wmr928eval (07.11.2009 03:10:02): error while opening file /data/weather//hist-last24h for writing: Input/output error: Input/output error

So I ssh to the SLUG, and try to check the /data/weather directory. I get odd output:

root@meteohub:/data/weather$ ls -l
ls: seq-month1: Input/output error
ls: hist-day1: Input/output error
ls: hist-year1: Input/output error

... (more Input/output errors for different files)

ls: thb1-month1: Input/output error
ls: 200904: Input/output error
total 20
drwxr-xr-x 2 root root 4096 Nov 7 2009 200811
drwxr-xr-x 2 root root 4096 Nov 6 17:47 200901
drwxr-xr-x 2 root root 4096 Apr 30 2009 200905
drwxr-xr-x 2 root root 4096 May 31 17:00 200906
-rw-r--r-- 1 root root 627 Nov 7 2009 hist-alltime
root@meteohub:/data/weather$

So it appears there is something VERY messed up about the file system on the SLUG. I can't manipulate any of the files in /data/weather using Linux file commands. For instance:

root@meteohub:/data/weather$ rm -rf 200811
rm: cannot remove directory `200811': Read-only file system

Or, even better:

root@meteohub:/data/weather$ rm -rf 200811
rm: cannot remove directory `200811': Read-only file system

I have reached wit's end. Any help will be appreciated.

-Bill

[skyewright]

and re-building the 2 GB USB flash drive (using the 4.5 firmware). The problem remains.

Have you tried using a different USB flash drive?

The number one source of problems with NSLU2 based Meteohubs is failure of the USB stick. :(

[bmoffitt]

I haven't, because the "stick" reads just fine when I plug it into my Linux box. No problems arise when I dd'ed the image to it, and the two partitions on the disk looked just fine when it was done.

That said, the top light on the SLUG is currently blinking, which (it says elsewhere) is indicative of a bad USB stick.

I have a spare 4G stick, so I'll try it.

Thanks,

Bill

[d_l]

The top LED on the NSLU2 can be blinking rapidly if there is a lot of activity to the USB stick and the meteohub and stick are both just fine when that happens. My top LED mostly blinks in unison with the USB stick's LED, but not always.

[bmoffitt]

OK, the USB stick seems to be the culprit. Finally got a new one built (BTW, whose idea was it to put the stick image, which is most easily transferred in Linux, in a .rar file, which I can't open in Linux???) and it seems to be working fine.

Weird...

Thanks!

-Bill

[mvpel]

I've found that even with USB sticks that claim to have error correction, they eventually degrade and lead to data corruption and most recently, failure of the unit. I set up something to run an fsck -c on the data volume once a week, but unfortunately you can't do that with the root filesystem and that's where the recent errors have occurred. I'm going to have to reflash I guess.

[bmoffitt]

I think you're right. At this point I suggest finding a USB stick that works well and buying a handful of them...