You are here: Home » Topic » 20060317 something gets stuck

20060317 something gets stuck

This topic contains 9 replies, has 2 voices, and was last updated by  rpedde 11 years, 7 months ago.

Viewing 10 posts - 1 through 10 (of 10 total)
  • Author
    Posts
  • #202

    tri
    Participant

    I got some weird effects with 20060317 in Linux.

    I was able to convert my configuration with this one and also
    file scan went ok (I made it from scratch). I can browse and play
    music fine with my Roku Soundbridge but after the server was up
    for about couple of hours, it got stuck. Server sockets were still
    open but inresponsive.

    If I sent kill to mt-daapd process, one of the threads didn’t die
    (of course kill -9 did the trick). However when I started a new
    mt-daapd process, it made the initial scan ok, but then immediately
    went to catatonic state. Maybe something was wrong in the database.

    I nuked the database and started it with mt-daapd -s and commended
    the scan from the web interface. Now it seems to work again. I’ll
    report again if it later gets stuck again.

    –tri

    #4102

    tri
    Participant

    Right, stuck again.

    This is how it manifests itself with me.

    % mt-daapd -s

    – Initiate scan from the web interface.

    – Use the server with Roku while it’s scanning

    – Everything seems to work and more and more songs appear

    – Scan completes

    – Try to browse artists from Roku

    – Browse timeouts, streaming that is going on continues ok.

    – Otherwise mt-daapd is inresponsive (web interface and
    new streaming is dead).

    I’ll try again by doing the initial database scan in foreground
    (i.e. without -s).

    –tri

    #4103

    tri
    Participant

    Nope, running the scan in foreground doesn’t help.
    Once the scan is done and service started trying to
    browse artists with Roku Soundbridge causes service
    thread to get stuck.

    After this pressing “reload” in web browser showing
    the status page, always allocates new thread session
    (i.e. new line appears to list of threads in the
    status page with each “reload” and old ones stay in
    the list).

    So something is wrong. I’ll try once more with
    iTunes.

    –tri

    #4104

    tri
    Participant

    Well, I wasn’t able to browse with iTunes either.
    With iTunes (6.0.2) I got “shared library not accessible
    for unknown reason (-3250)”. But after this mt-daapd doesn’t
    get stuck and web interface keeps working.

    Maybe the difference is that iTunes shuts down the connection
    bluntly but Roku tries to shut it down nicely and keeps
    hanging. This is however just a wind guess.

    Anyways, something is wrong.

    The weird thing is that Roku works nicely while the initial
    is going on if the server is started with -s and scan started
    via web interface.

    –tri

    #4105

    tri
    Participant

    I rolled back to 20060307. That is the last one working propely for me.

    –tri

    #4106

    rpedde
    Participant

    Yeah, this is from folding the browse/query stuff into the playlist parser. The interface that the soundbridge uses is completely different from the one iTunes uses. I jettisoned a bunch of query parsing code and tried to use the playlist parser to parse the soundbridge queries. Clearly it didn’t fit well, but once it actually works, it will be a net plus — both types of queries will have the same parse tree, and can be merged easily.

    Where I hope to go with this is to set a “base playlist” based on logon or password or connection ip address. Then you could use a smart playlist as your base library. So you could have your kids connect with a password and have their base playlist be ‘content_rating != “explicit”‘, and have smart playlists (and even static playlists) generated with respect to that as a base.

    That’s why I want to merge the parsers. Clearly it’s not quite there yet, though.

    — Ron

    #4107

    rpedde
    Participant

    a run at “-d10 -D parse” would help.

    — Ron

    #4108

    tri
    Participant

    Actually I got all new problem with 20060307 too that in the light of
    your comment is totally unrelated.

    When I went to sleep, I left two soundbridges streaming the playlist
    (different one for both) and the web browser to poll the status display.
    After a while (an hour or so) syslog bluntly tells:

    Mar 19 05:16:59 rinne mt-daapd[8971]: Could not spawn thread: Resource temporarily unavailable
    Mar 19 05:16:59 rinne mt-daapd[8971]: Fatal error… Aborting
    Mar 19 05:16:59 rinne mt-daapd[8967]: Rendezvous socket closed (daap server crashed?) Aborting.
    Mar 19 05:16:59 rinne mt-daapd[8967]: Fatal error… Aborting

    –tri

    #4109

    tri
    Participant

    rpedde wrote:

    a run at “-d10 -D parse” would help.

    I only got this:

    633.rinne:src> root ./mt-daapd -f -c /etc/mt-daapd.conf.new -d10 -D parse
    Adding module parse to debug list (0x00002000)
    Debug mask is 0x80002000

    – In this phase I connected with Roku and got “Failed to load browse data”

    – After this it was stuck and I pressed Control-C and got:

    Segmentation fault
    Exit 139 (SIGSEGV)

    Anyways, I think you got something in this fixed already because of
    your comment in the Roku thread.

    –tri

    #4110

    rpedde
    Participant

    Yeah, I’ll go ahead and drop this tonight.

Viewing 10 posts - 1 through 10 (of 10 total)

You must be logged in to reply to this topic.