pablo,
I am sitting in the Chicago airport and reloading build 15 right now VNCserver is a great tool. I should know in about an hour or two if it has the same issue. If I find out before I fly out I will respond here.
Mike - N7DQ
pablo,
I am sitting in the Chicago airport and reloading build 15 right now VNCserver is a great tool. I should know in about an hour or two if it has the same issue. If I find out before I fly out I will respond here.
Mike - N7DQ
Hi Mike,
Have a safe flight.
pablo,
Same issue with build 15. About an hour into the run I get the first zombie/defunct process. So I started the process at 15:29 local and at 16:20 local got a defunct process. Can’t tell what caused it though.
Mike - N7DQ
Hi Mike,
My guess is that the parent process is fork’ing the child and the child dies. The parent doesn’t catch death of a child' and hence, the zombies. This is why
nohup’ isn’t working either. `nohup’ would work if you had a single process.
In short, this is a problem for Brian to fix.
pablo,
I am wondering if this has to do with the cronftp process. At some point earlier this week, I know I had well over 10 zombies/defunct and my server stopped doing ftp for quite some time. I am writing a perl script to tail the log file so if it happens again some test will be run to find out if it is my server, router or network provider is causing the problem. I do not think it was the network provider as the outage was almost 16 hours and I think maybe my server was having problems cleaning up broken ftp connections.
Any way just some speculation.
Mike - N7DQ
Hi Mike,
There definitely was a problem with cronftp' and their brethren. In your WD directory, is there a subdirectory named
lib’ And is there a library named `libutils’? This is the library I wrote (as I mentioned previously, I snarfed most of the code off the web).
it will be the problem with the rainrecorded.wav file trying to be played but you have no sound setup
delete that file, so that is not played
(from the other thread about that)
Why not use the fire_and_forget() function? That’d avoid the zombie issues once and for all.
I removed the rainrecorded.wav file (reappeared when I dropped back to build 15) and reloaded build 16. I am seeing in about 15 hours 8 zombies still. I am going to let things run to see if I get any ftp problems which it seemed like happened early in the week.
I am seeing access violations and wonder if these might be time related - start at around mid-night and then they start again around 3am. I am trying to help get data for you.
Mike - N7DQ
Ok another update.
At 7:55am I had 8 zombies and at 8:05am there were 0 zombies. The 8 zombies started about 1:00am. It took about 12 hours to reach the total of 8 zombies and about 4 hours for the first zombie to appear.
Not sure this helps.
Mike - N7DQ
have you been getting rain?
Another problem has come that maybe related I am have trouble with crashing. in the error file I am getting" not able to write to WDISPLAY.INI " ( permissions are correct)It get about four access violations just before . It has been running in the past for Months without crashing.
Brian,
I have not been getting rain at least back to about 12/5ish (snow).
Mike
Brian,
Looks like I might have gotten rain in the form f melted frost that registered today, if that helps.
Mike - N7DQ
One thing I have notice consistently is that at some time between 7:55am and 8am local a process cleans up some of the defunct processes. I am not running anything, what is WD running?
I have a process that counts the WeatherD processes and counts the defunct processes and finally grep and count for any access violations in the log file. The format is
date/time - WD processes:Defunct WD process:Access violation errors in the log
200812190733 - 010:009:0
200812190738 - 011:009:0
200812190743 - 012:010:0
200812190748 - 011:010:0
200812190753 - 012:010:0
200812190758 - 001:000:0
200812190803 - 002:000:0
Mike - N7DQ
I built a perl script that grabs the date/time, total number of WeatherD process, total number of defunct WeatherD process and the number of access violations in the log file. It updates the page every 5 minutes (multiples of 3 and 8 ).
http://www.n7dq.com/weather/WD-Status2.html Note: You can find my start time and data packets received at Island Drive Weather - Longview, WA - Home
It gives you and idea of how these processes are building. If anyone wants the source let me know.
Mike - N7DQ