[mythtv-commits] Ticket #10911: another backend deadlock
MythTV
noreply at mythtv.org
Mon Jul 16 15:04:51 UTC 2012
#10911: another backend deadlock
----------------------------------------+----------------------------
Reporter: brian@… | Owner:
Type: Bug Report - Hang/Deadlock | Status: new
Priority: minor | Milestone: unknown
Component: MythTV - General | Version: 0.25-fixes
Severity: medium | Resolution:
Keywords: | Ticket locked: 0
----------------------------------------+----------------------------
Comment (by newbury@…):
Commenting from work. I am also getting the dreaded MYTH_PROTO_VERSION is
empty message from mythfrontend.
Repeatable, and instant. Backend cannot record. This is MORE than a minor
fault. Evidenced in a trunk build from about a week ago, Rebuilt from
master trunk from about 11:00 pm EDT last night, and compiled early this
morning. Single box BE/FE, Fedora 17. Updated to schema 1307 on first
launch this morning.
What is happening may provide a clue for the spelunking. I launch
mythfrontend and it comes up cleanly. I can then: Open Watch Recordings,
and see the list of programs in mythconverg.recorded. I can play any of
them. I can exit to Schedule Recordings, and enter the Program Guide, or
the Program Finder, or do any keyword search.
But in Upcoming Recordings, I get "You have not scheduled..." And in the
console I see the dreaded error. Meanwhile I can open a mysql console as
user mythtv etc, and see that table mythconverg.record has 111 records.
If I try a Manual Schedule, it *appears* to run, but produces no file (not
even a zero byte file). I get a 'file cannot be found' error when I select
the recording in Watch Recordings. That is, it creates the 'recorded'
record but the tuner side is 'missing in in-action'.
I also cannot get anything in Watch TV.. It hangs on 'Please Wait' whereas
it should give me the EPG.
Restart of the backend makes no difference. Same 'empty' error in the
console every time. More than 'repeatable'. I googled the error and
checked the possible error conditions. One post on how to fix the the
usual causes of this error was one of my own from 2010! And none of those
usual conditions applies. I optimize'd on every relaunch cycle in case a
file/table was being corrupted. No joy.
So this is either two distinct deadlock conditions or one deadlock found
in the common code for the 'Upcoming Recordings' retrieval code and in the
scheduling code. How those two threads would cross paths is a little
unclear to me, but I am assuming there is a common base, deep in the code
in a section which deals with 'and now talk to the database'. It is must
be tied to the scheduling code as the EPG and Finder and Watch Recordings
sections work properly.
So I cannot record ATM. The WAF will PLUNGE.. This is more than 'babysat
daily'.
What would you like to see in the way of logs, etc?
I am going to have to rebuild. Would it help to dissect if I created a new
branch from git, say 2/3/4 weeks ago, and built from that?
Geoff Newbury not logged in... TRAC doesn't seem to like my password any
more...
--
Ticket URL: <http://code.mythtv.org/trac/ticket/10911#comment:5>
MythTV <http://code.mythtv.org/trac>
MythTV Media Center
More information about the mythtv-commits
mailing list