- May 28, 2011
William G. is on it, that is a good thing. I'm certain that once it works properly that he will create an updated MiniDLNA plugin.
I'm going to try with a small test dir and see if I can keep it running..[2014/07/03 19:40:50] inotify.c:857: warn: kevent polling failure: Interrupted s ystem call
[2014/07/03 19:41:27] upnphttp.c:1890: info: Serving DetailID: 192 [/tmp/Gumball .mp4]
[2014/07/03 19:41:27] inotify.c:857: warn: kevent polling failure: Interrupted s ystem call
[2014/07/03 19:44:26] minidlna.c:1210: error: select(all): Bad file descriptor
[2014/07/03 19:44:26] minidlna.c:1211: fatal: Failed to select open sockets. EXI TING
Oddly leaving clients running and re-launching minidlna causes the same issue immediately.. I think my setup is going to require alot of testing to narrow down the issues.. I think however this 113.11 should be the new testing version go forward..[2014/07/04 21:41:34] upnpevents.c:423: fatal: upnpevents_selectfds: file descriptor 8187 too big for select, limit is 1024
First, thanks for your diligence in locating the information.Looks like the kqueue with large libraries still causes issues here.. Sockets appeared to be one time issue probably from the crash from upnpevents.c:423: fatal: upnpevents_selectfds: file descriptor 8187 too big for select, limit is 1024 ...
Any idea how to increase this limit?
I just found this but no idea if it will work.. https://email@example.com/msg59259.html
This is good news as we now have something to work with. I did ask him if he wanted any feedback because I don't want to force feedback on him if he doesn't want it.Hi Joe,
I have made the patch available at:
(to be applied with patch -p from within "/usr/ports/net/minidlna"
or without -p in the "files" subdirectory.)
This patch is what I'm using. It is not cleaned-up with regard to
(1) indentation and (2) reduction to functional changes only.
The reason for (1) is that I used diff with options to ignore white-space changes. Else, the patch would be much larger, due to indentation changes for large parts of the code, which is unmodified, else. The reason for (2) is that I added quite a number of debug messages for my tests and have not yet decided, which ones to keep or to remove. But these statements will be ignored, if you do not start minidlna with the "-d" or "-v" options, and they do not add much to code size or execution time.
> I am curious if you would provide further information on your changes
> and also provide a copy of a compiled/installable version for testing?
> I haven’t compiled anything in literally decades which is why I’d also
> like a compiled copy.
The compiled package (PKG_NG style) is now available from:
This is the exact binary I'm currently using, built under -CURRENT.
It may just work on FreeBSD-9 (amd64), but you may need to extract manually extract it from the package file.
I understand, that this is not what you need for FreeNAS, but I do not know whether you want an *i386* or *amd64* FreeBSD-9 binary.
(I could build it for you in a poudriere jail, if you'd specify what you need ...)
Best regards, STefan