[mythtv-users] oom-killer on the loose.

[Richard Thomas]

Michael T. Dean wrote:
> And switching to Schedulers Direct (for only $0.055 per day) will likely 
> fix the problem a /lot/ quicker than the devs (likewise not picking on 
> the devs, but meaning that a lot of time has already been spent on 
> finding out why there's a memory leak and why it only affects some 
> people--and the leak itself still has to be found).
>   
It would be nice if that would fix it. OTOH, it would also be nice to 
know what the issue actually was. Is it definitely an EIT thing? I'm 
definitely going SD anyway, I just need to schedule a couple of hours so 
that if things go wrong, there's time to fix it :)
> Oh, and if you think about it--let's say that it takes devs 70 days*** 
> of work to track down the issue.  That would only be $3.84 worth of a SD 
> subscription.  Or, if you do the 1-week free trial, and /then/ 
> subscribe, it's only $3.45 (yes, you can save almost $0.39 by signing up 
> for that free one-week trial).  And, in the meantime, you'd actually 
> have excellent data that makes you question your having commented that 
> EIT in your are "work[s] pretty well even if it is a little short on 
> data." ;)
>   
The EIT was more of a "let's give it a shot". It actually has a lot of 
potential. A couple of the channels have 2 weeks of guide data. 
Unfortunately, the other channels maybe have 12 hours and are incorrect 
in some places. Being incorrect is a little surprising considering that 
the source is from the channels directly. On the other hand, maybe it is 
also a symptom of whatever bug is causing the OOM issues.
> If that's not a good enough plan for you, you could try to buy some more 
> RAM.  Let's say that you spend the $20 to buy 4GB more RAM.  That may 
> allow you to run Myth for 4-8 days before the system locks up.
>   
More swap would be a more likely solution. Given that it only happens 
every few days, it seems possible that I am "only just" running out of 
memory and a couple of more gigs of swap would solve that neatly. Though 
as I say, I really haven't seen any memory issues directly so far. I'll 
try some monitoring and see what pops up.
> Or you could hire a developer (what's the standard rate?  $100/hr?) to 
> fix it for us.  :)
>   
I am a developer (C/C++, perl, php, whatever) and I'm not above getting 
my hands dirty on this. mythtv seems like something worth donating some 
time to. OTOH, if someone already knows a solution or can say "run this 
and it will tell you where to look for the problem", it would be a lot 
easier. Sounds like I need to look into this valgrind thing.

Do you have a couple of ticket numbers for those issues? (I can probably 
find them with a search but a direct reference always speeds things along).
> Mike
>
> ***70 is just a number I happened to type.  Feel free to interpret that 
> to mean "it may be fixed within 70 days" or "the devs may spend 70 
> "business" (8-hour) days of their available MythTV development time 
> working the issue" or "the devs may spend 70 full (24-hours) days of 
> their Myth dev time working the issue".  Your guess about when it will 
> be fixed/how long it will take is as good as mine (feel free to consult 
> Trac to see just how long the tickets for it have been open/how long 
> they've already been working).  The number here is actually meaningless, 
> but just used to indicate that you get a lot for your $20 for an SD 
> subscription.  :)
> _______________________________________________
> mythtv-users mailing list
> mythtv-users at mythtv.org
> http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users
>
>