Headless content server confusion

[darklajid]

Hey there.

I recently installed calibre (1.13) as a headless content server: Just moved my library to a machine that's always on, installed it completely standalone using

sudo python -c "import sys; py3 = sys.version_info[0] > 2; u = __import__('urllib.request' if py3 else 'urllib', fromlist=1); exec(u.urlopen('http://status.calibre-ebook.com/linux_installer').read()); main()"

This worked for a time. Now I'm in the weird situation that it works _sometimes_. Most of the time starting calibre-server leads to these lines and the process dieing:

Dec 08 12:35:57 bitdump calibre-server[1243]: Traceback (most recent call last):
Dec 08 12:35:57 bitdump calibre-server[1243]: File "site-packages/calibre/library/server/base.py", line 272, in start
Dec 08 12:35:57 bitdump calibre-server[1243]: File "site-packages/calibre/library/server/base.py", line 265, in start_cherrypy
Dec 08 12:35:57 bitdump calibre-server[1243]: File "site-packages/cherrypy/process/wspbus.py", line 250, in start
Dec 08 12:35:57 bitdump calibre-server[1243]: ChannelFailures: IOError(2, 'No such file or directory')

There's nothing special about the way I launch the thing:

/usr/bin/calibre-server --max-cover=600x800 --max-opds-items=30 --max-opds-ungrouped-items=100 --username=calibre --port=8888 --pidfile=/var/run/calibre/calibre-server.pid --with-library=/tank/books

The thing is: Sometimes .. it works/stays up. I tried following the stack trace, but I don't see anything special in there, certainly nothing about potential files/directories. I'd be glad if someone could help me fix this situation.

Minor sidenote, added but low priority issue: I cannot use --daemonize with calibre-server, which is the reason why that option is absent from my line above. I'd prefer to use that, but ..

[bitdump dar]# /usr/bin/calibre-server --max-cover=600x800 --max-opds-items=30 --max-opds-ungrouped-items=100 --username=calibre --port=8888 --pidfile=/var/run/calibre/calibre-server.pid --with-library=/tank/books --daemonize
Aborted (core dumped)

[kovidgoyal]

Most likely that's because something in your environment is different each time. Check the environment variables, check that the directory containing the pid file exists and so on. And if --daemonize is causing a core dump on your server, it points to something fairly serious being messed up, as all --daemonize does is this,

def daemonize(stdin='/dev/null', stdout='/dev/null', stderr='/dev/null'):
    try:
        pid = os.fork()
        if pid > 0:
            # exit first parent
            sys.exit(0)
    except OSError as e:
        print >>sys.stderr, "fork #1 failed: %d (%s)" % (e.errno, e.strerror)
        sys.exit(1)

    # decouple from parent environment
    os.chdir("/")
    os.setsid()
    os.umask(0)

    # do second fork
    try:
        pid = os.fork()
        if pid > 0:
            # exit from second parent
            sys.exit(0)
    except OSError as e:
        print >>sys.stderr, "fork #2 failed: %d (%s)" % (e.errno, e.strerror)
        sys.exit(1)

    # Redirect standard file descriptors.
    si = file(stdin, 'r')
    so = file(stdout, 'a+')
    se = file(stderr, 'a+', 0)
    os.dup2(si.fileno(), sys.stdin.fileno())
    os.dup2(so.fileno(), sys.stdout.fileno())
    os.dup2(se.fileno(), sys.stderr.fileno())

where i dont see any potential for SIGABRT unless some system call is failing.

[darklajid]

Hey.

First of all: Thanks for the feedback and your great work on calibre in general.
Two things: I guess you helped me track down the startup issue - it's probably (still investigating) related to /var/run being cleaned after a reboot. So .. thanks and sorry for the noise..

Regarding the core dump: The machine works fine. The only thing I can think of as worth mentioning is that this is a Arch Linux box, running python3 by default (not sure if that might cause trouble). As far as I understood the installation procedure though everything related to calibre should be below /opt now, independent of my system's state of affair?

The coredump/backtrace is probably useless?

(gdb) bt
#0  0x00007f329afa83d9 in raise () from /usr/lib/libc.so.6
#1  0x00007f329afa97d8 in abort () from /usr/lib/libc.so.6
#2  0x00007f32996cb3d3 in ffi_closure_free () from /opt/calibre/lib/libffi.so.4
#3  0x00007f32998df99f in ?? () from /opt/calibre/lib/python2.7/lib-dynload/_ctypes.so
#4  0x00007f329ac36ddb in ?? () from /opt/calibre/lib/libpython2.7.so.1.0
#5  0x00007f32998d88b7 in ?? () from /opt/calibre/lib/python2.7/lib-dynload/_ctypes.so
#6  0x00007f32998d8aa9 in ?? () from /opt/calibre/lib/python2.7/lib-dynload/_ctypes.so
#7  0x00007f329ac5997d in ?? () from /opt/calibre/lib/libpython2.7.so.1.0
#8  0x00007f329ac35183 in ?? () from /opt/calibre/lib/libpython2.7.so.1.0
#9  0x00007f329ac37b30 in ?? () from /opt/calibre/lib/libpython2.7.so.1.0
#10 0x00007f329ac3aaad in _PyModule_Clear () from /opt/calibre/lib/libpython2.7.so.1.0
#11 0x00007f329acadbf1 in PyImport_Cleanup () from /opt/calibre/lib/libpython2.7.so.1.0
#12 0x00007f329acb73db in Py_Finalize () from /opt/calibre/lib/libpython2.7.so.1.0
#13 0x00007f329b3203f5 in execute_python_entrypoint ()
   from /opt/calibre/lib/libcalibre-launcher.so
#14 0x00000000004006ff in main ()

[kovidgoyal]

Yes the binary calibre install does not depend on the system python in anyway.