Mailing List Archive: MythTV: Users

mythbackend still eats memory: the current status

1 2 3 4 5 6 7

View All

Index | Next | Previous | View Threaded

william_munson at comcast

May 22, 2009, 1:31 PM

Post #101 of 155 (3003 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 05/20/2009 08:24 AM, Udo van den Heuvel wrote:
> Alex Butcher wrote:
>> My hardware is a P4 2.53GHz with 768MB of RAM, a Hauppauge Nova-T and a
>> Nova-T-500 (so three DVB-T physical tuners altogether). I use
>> multirec to
>> provide two virtual tuners per physical tuner. The UK has 6
>> multiplexes. I
>> probably record at least 4 hours per day most days. EIT is enabled. The
>> kernel is 2.6.26.6-49.fc8 using the standard cx88_dvb and
>> dvb_usb_dib0700
>> drivers supplied by that kernel.
>
> Thanks.
> This somehow, perhaps even undeniably, shows that `the problem` is not
> just limited to my mythTV configuration, my build, my hardware or my
> distribution.
>
> Udo

I am also see the same memory leak issue but because my machine is no
where's near powerful enough to valgrind I cannot provide any useful
feedback myself. For me, the fix so far has been several scripts
monitoring myth and rebooting, resetting and cleaning up as needed. Not
100% but its much better now. A real fix would be nice.

_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

May 24, 2009, 12:16 AM

Post #102 of 155 (2981 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

William wrote:
>> Thanks.
>> This somehow, perhaps even undeniably, shows that `the problem` is not
>> just limited to my mythTV configuration, my build, my hardware or my
>> distribution.
>>
> I am also see the same memory leak issue but because my machine is no
> where's near powerful enough to valgrind I cannot provide any useful
> feedback myself. For me, the fix so far has been several scripts
> monitoring myth and rebooting, resetting and cleaning up as needed. Not
> 100% but its much better now. A real fix would be nice.

Could you explain what software version you run, what your typical usage
pattern is and what memory the backend consumes over time?

Kind regards,
Udo

%CPU %MEM VSZ RSS TTY STAT START TIME
11.0 8.7 407028 85280 ? Ssl 03:06 12:59
10.9 11.6 445264 113104 ? Ssl May06 170:32
10.8 14.4 474364 140528 ? Ssl May06 325:50
10.8 15.4 482604 149604 ? Ssl May06 483:36
10.7 16.3 492400 158516 ? Ssl May06 632:38
10.6 17.3 501680 168732 ? Ssl May06 781:28
10.7 18.8 515120 182808 ? Ssl May06 940:08
10.8 20.7 534032 201016 ? Ssl May06 1105:05
10.8 22.0 546336 213748 ? Ssl May06 1266:19
10.9 23.3 559368 226224 ? Ssl May06 1429:23
10.9 24.9 575364 242112 ? Ssl May06 1591:37
10.9 25.7 575424 250272 ? Ssl May06 1742:04
10.8 25.7 592900 250248 ? Ssl May06 1895:01
10.9 26.8 605720 260412 ? Ssl May06 2058:53
10.9 27.2 618236 264712 ? Ssl May06 2224:00
10.9 27.4 630772 266556 ? Ssl May06 2386:19
10.9 28.0 647244 272596 ? Ssl May06 2547:03
11.0 28.6 680556 278380 ? Ssl May06 2710:26
10.9 28.7 690352 279048 ? Ssl May06 2861:30

_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

May 29, 2009, 8:02 AM

Post #103 of 155 (2921 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

David Lister wrote:
> your graphs. I don't have problems personally and I can't measure or
> audit anything, because I'm on a project in India. It's 6pm, so see you
> tomorrow!

So were you able to look at my *numbers* as opposed to the graphs?

Please let me know your conclusions, theories and opinions about them.
For your convenience I will attach them at the bottom.
FYI: They are collected at 24 hour intervals.

Kind regards,
Udo

%CPU %MEM VSZ RSS TTY STAT START TIME
11.0 8.7 407028 85280 ? Ssl 03:06 12:59
10.9 11.6 445264 113104 ? Ssl May06 170:32
10.8 14.4 474364 140528 ? Ssl May06 325:50
10.8 15.4 482604 149604 ? Ssl May06 483:36
10.7 16.3 492400 158516 ? Ssl May06 632:38
10.6 17.3 501680 168732 ? Ssl May06 781:28
10.7 18.8 515120 182808 ? Ssl May06 940:08
10.8 20.7 534032 201016 ? Ssl May06 1105:05
10.8 22.0 546336 213748 ? Ssl May06 1266:19
10.9 23.3 559368 226224 ? Ssl May06 1429:23
10.9 24.9 575364 242112 ? Ssl May06 1591:37
10.9 25.7 575424 250272 ? Ssl May06 1742:04
10.8 25.7 592900 250248 ? Ssl May06 1895:01
10.9 26.8 605720 260412 ? Ssl May06 2058:53
10.9 27.2 618236 264712 ? Ssl May06 2224:00
10.9 27.4 630772 266556 ? Ssl May06 2386:19
10.9 28.0 647244 272596 ? Ssl May06 2547:03
11.0 28.6 680556 278380 ? Ssl May06 2710:26
10.9 28.7 690352 279048 ? Ssl May06 2861:30
10.9 29.1 691436 282472 ? Ssl May06 3011:50
10.9 29.2 711492 284056 ? Ssl May06 3174:03
10.9 29.2 725916 283748 ? Ssl May06 3335:27
10.9 29.4 735728 285392 ? Ssl May06 3494:58
10.9 29.8 749644 289800 ? Ssl May06 3653:39

_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Jun 27, 2009, 7:47 AM

Post #104 of 155 (2817 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2009-05-29 17:02, Udo van den Heuvel wrote:
> David Lister wrote:
>> your graphs. I don't have problems personally and I can't measure or
>> audit anything, because I'm on a project in India. It's 6pm, so see you
>> tomorrow!
>
> So were you able to look at my *numbers* as opposed to the graphs?
>
> Please let me know your conclusions, theories and opinions about them.
> For your convenience I will attach them at the bottom.
> FYI: They are collected at 24 hour intervals.

Since there was no real reply, here is *my* graph, made from *my*
numbers which you find at the botom:

http://img151.imageshack.us/img151/4247/mythbackendeatsmemory.png

It clearly shows at least one trend: a growing VSZ.
Maybe you can explain why I don't have to worry.

Kind regards,
Udo

%CPU %MEM VSZ RSS TTY STAT START TIME
10.9 11.6 445264 113104 ? Ssl May06 170:32
10.8 14.4 474364 140528 ? Ssl May06 325:50
10.8 15.4 482604 149604 ? Ssl May06 483:36
10.7 16.3 492400 158516 ? Ssl May06 632:38
10.6 17.3 501680 168732 ? Ssl May06 781:28
10.7 18.8 515120 182808 ? Ssl May06 940:08
10.8 20.7 534032 201016 ? Ssl May06 1105:05
10.8 22.0 546336 213748 ? Ssl May06 1266:19
10.9 23.3 559368 226224 ? Ssl May06 1429:23
10.9 24.9 575364 242112 ? Ssl May06 1591:37
10.9 25.7 575424 250272 ? Ssl May06 1742:04
10.8 25.7 592900 250248 ? Ssl May06 1895:01
10.9 26.8 605720 260412 ? Ssl May06 2058:53
10.9 27.2 618236 264712 ? Ssl May06 2224:00
10.9 27.4 630772 266556 ? Ssl May06 2386:19
10.9 28.0 647244 272596 ? Ssl May06 2547:03
11.0 28.6 680556 278380 ? Ssl May06 2710:26
10.9 28.7 690352 279048 ? Ssl May06 2861:30
10.9 29.1 691436 282472 ? Ssl May06 3011:50
10.9 29.2 711492 284056 ? Ssl May06 3174:03
10.9 29.2 725916 283748 ? Ssl May06 3335:27
10.9 29.4 735728 285392 ? Ssl May06 3494:58
10.9 29.8 749644 289800 ? Ssl May06 3653:39
10.9 30.9 762536 300424 ? Ssl May06 3814:24
10.9 30.9 774420 300832 ? Ssl May06 3963:43
10.9 27.6 783192 268840 ? Ssl May06 4112:59
10.9 27.2 792472 264128 ? Ssl May06 4261:44
10.9 26.9 773596 261840 ? Ssl May06 4421:17
10.9 27.8 814144 270620 ? Ssl May06 4580:04
10.9 27.3 825384 265892 ? Ssl May06 4735:53
10.9 26.2 836144 254296 ? Ssl May06 4894:31
10.9 26.5 843884 257444 ? Ssl May06 5040:09
10.8 26.7 852656 260060 ? Ssl May06 5182:13
10.8 27.7 863844 269792 ? Ssl May06 5333:51
10.8 28.7 876220 279444 ? Ssl May06 5486:34
10.8 29.8 888596 290100 ? Ssl May06 5644:06
10.8 31.1 905200 302120 ? Ssl May06 5801:33
10.8 31.8 917784 309512 ? Ssl May06 5960:15
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Mar 23, 2010, 7:08 AM

Post #105 of 155 (2507 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

Hello,

As you may have noticed, mythtv 0.22-fixes is now running on a Jetway
NC81-LF with AMD 235e CPU, Fedora 12 x86_64.

Despite the differences with the Epia system we see a similar trend:

%CPU %MEM VSZ RSS T
6.1 2.4 1683620 94648
5.4 2.6 1709504 100304
5.7 2.9 1709504 113244
5.6 3.5 1709504 133392
5.6 3.9 1709504 149016
5.6 4.3 1709504 165208

Mythbackend runs solidly so far, despite the RSS growth
I'll try to fix the audio issues for the frontend this weekend and after
that we may be able to start the valgrind sessions...

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Mar 28, 2010, 2:09 AM

Post #106 of 155 (2452 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-03-23 15:08, Udo van den Heuvel wrote:
> Mythbackend runs solidly so far, despite the RSS growth
> I'll try to fix the audio issues for the frontend this weekend and after
> that we may be able to start the valgrind sessions...

The HDMi audio is fixed
Next will be the valgrind run to find out about the RSS growth.

What would be the optimal way to start valgrind?
Any options I should specify?

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

nick.rout at gmail

Mar 28, 2010, 2:28 AM

Post #107 of 155 (2430 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On Sun, Mar 28, 2010 at 10:09 PM, Udo van den Heuvel <udovdh [at] xs4all> wrote:
> On 2010-03-23 15:08, Udo van den Heuvel wrote:
>> Mythbackend runs solidly so far, despite the RSS growth
>> I'll try to fix the audio issues for the frontend this weekend and after
>> that we may be able to start the valgrind sessions...
>
> The HDMi audio is fixed
> Next will be the valgrind run to find out about the RSS growth.
>
> What would be the optimal way to start valgrind?
> Any options I should specify?

google search term valgrind mythtv finds:

http://www.gossamer-threads.com/lists/mythtv/dev/300776#300776
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Mar 28, 2010, 2:57 AM

Post #108 of 155 (2433 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-03-28 11:28, Nick Rout wrote:
>> The HDMi audio is fixed
>> Next will be the valgrind run to find out about the RSS growth.
>>
>> What would be the optimal way to start valgrind?
>> Any options I should specify?
>
> google search term valgrind mythtv finds:
>
> http://www.gossamer-threads.com/lists/mythtv/dev/300776#300776

Ah, so `valgrind --tool=memcheck --error-limit=no --leak-check=full
--log-file=backend-valgrind.log -v -- /usr/local/bin/mythbackend | tee
backend.log` is also especially appropriate for this case.

Thanks!

Kind regards,
Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Mar 31, 2010, 6:08 AM

Post #109 of 155 (2355 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-03-28 11:57, Udo van den Heuvel wrote:
>> http://www.gossamer-threads.com/lists/mythtv/dev/300776#300776
>
> Ah, so `valgrind --tool=memcheck --error-limit=no --leak-check=full
> --log-file=backend-valgrind.log -v -- /usr/local/bin/mythbackend | tee
> backend.log` is also especially appropriate for this case.

At last, I now have a valgrind session running. The backend is working
so all is well.
I now have a process like this:

valgrind --tool=memcheck --error-limit=no --leak-check=full
--log-file=/home/mythtv/backend-valgrind.log -v -- /usr/bin/mythbackend
--logfile /var/log/mythtv/mythbackend.log --pidfile /var/run/mythbackend.pid

Can I just kill the valgrind thing and hope it logs the results as it
ends? (I intend to do so this next weekend)

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Mar 31, 2010, 6:32 AM

Post #110 of 155 (2350 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-03-31 15:08, Udo van den Heuvel wrote:
> Can I just kill the valgrind thing and hope it logs the results as it
> ends? (I intend to do so this next weekend)

Never mind.
I noticed that not all was well so I ctrl-C'd the small script that I
used to start the backend.
I now have some logging; approximately 1325689 bytes.
What should be in there if we really have found an issue?

Udo

_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Mar 31, 2010, 6:38 AM

Post #111 of 155 (2354 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-03-28 11:57, Udo van den Heuvel wrote:
>> http://www.gossamer-threads.com/lists/mythtv/dev/300776#300776
>
> Ah, so `valgrind --tool=memcheck --error-limit=no --leak-check=full
> --log-file=backend-valgrind.log -v -- /usr/local/bin/mythbackend | tee
> backend.log` is also especially appropriate for this case.

Well, this gives me logging like:

2010-03-31 15:37:15.925 DevRdB(/dev/dvb/adapter0/frontend0) Error:
Driver buffers overflowed
2010-03-31 15:37:21.062 DevRdB(/dev/dvb/adapter0/frontend0) Error:
Driver buffers overflowed

Any idea on how to fix these?

Kind regards,
Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Mar 31, 2010, 9:25 PM

Post #112 of 155 (2327 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-03-31 15:32, Udo van den Heuvel wrote:
> I now have some logging; approximately 1325689 bytes.
> What should be in there if we really have found an issue?

See http://svn.mythtv.org/trac/ticket/8266 for the ticket.

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Apr 1, 2010, 5:32 AM

Post #113 of 155 (2283 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-04-01 06:25, Udo van den Heuvel wrote:
> See http://svn.mythtv.org/trac/ticket/8266 for the ticket.

See the updated status.
As you will notice explanation is required, perhaps on the wiki, why
valgrinds 'definitely' is not so definitely according to other people.

I did add another log file with a short valgrind run with also a few
definitely's. So let's see what happens.

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Apr 1, 2010, 10:12 AM

Post #114 of 155 (2272 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-04-01 14:32, Udo van den Heuvel wrote:
> On 2010-04-01 06:25, Udo van den Heuvel wrote:
>> See http://svn.mythtv.org/trac/ticket/8266 for the ticket.
>
> See the updated status.

At least we see a start of fixing the leaks.

Can anybody explain how much RAM this one fix would cost (if left
unfixed) when recording 3 channels 24/7 off of one DVB-T multiplex?

I.o.w.: is this *the* issue or are there more leaks?

I'll see what 0.23-fixes does for leaks after the weekend.

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

mrand at pobox

Apr 1, 2010, 11:29 AM

Post #115 of 155 (2271 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On Thu, Apr 1, 2010 at 12:12 PM, Udo van den Heuvel <udovdh [at] xs4all> wrote:
> On 2010-04-01 14:32, Udo van den Heuvel wrote:
>> On 2010-04-01 06:25, Udo van den Heuvel wrote:
>>> See http://svn.mythtv.org/trac/ticket/8266 for the ticket.
>>
>
> At least we see a start of fixing the leaks.
>
> Can anybody explain how much RAM this one fix would cost (if left
> unfixed) when recording 3 channels 24/7 off of one DVB-T multiplex?
>
> I.o.w.: is this *the* issue or are there more leaks?
>
>
> I'll see what 0.23-fixes does for leaks after the weekend.

My limited understanding is that this might be 50-60 KBytes per
recording and that entering LiveTV might be double that.

Have fun,

Marc
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Apr 21, 2010, 6:34 AM

Post #116 of 155 (2060 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

http://svn.mythtv.org/trac/ticket/8266#comment:21

There's some progress!

Kind regards,
Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Apr 22, 2010, 7:34 AM

Post #117 of 155 (2046 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-04-21 15:34, Udo van den Heuvel wrote:
> http://svn.mythtv.org/trac/ticket/8266#comment:21
>
> There's some progress!

Some early figures:

%CPU %MEM VSZ RSS TTY STAT START TIME C
5.1 2.2 1706748 87208 ? Ssl Apr17 41:00 /usr/bin/myt
5.2 2.2 1706748 86804 ? Ssl Apr17 116:20 /usr/bin/myt
5.3 2.5 1706748 95612 ? Ssl Apr17 198:05 /usr/bin/myt
5.4 2.6 1706748 102480 ? Ssl Apr17 278:57 /usr/bin/myt
5.4 2.8 1706748 106516 ? Ssl Apr17 358:35 /usr/bin/myt

Indeed slightly weird but the trend is a slower RSS rise than before!!

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

johan.heikkila at gmail

Apr 22, 2010, 10:54 AM

Post #118 of 155 (2030 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

2010/4/22 Udo van den Heuvel <udovdh [at] xs4all>:
> On 2010-04-21 15:34, Udo van den Heuvel wrote:
>> http://svn.mythtv.org/trac/ticket/8266#comment:21
>>
>> There's some progress!
>
> Some early figures:
>
> %CPU %MEM � �VSZ � RSS TTY � � �STAT START � TIME C
> �5.1 �2.2 1706748 87208 ? � � � Ssl �Apr17 �41:00 /usr/bin/myt
> �5.2 �2.2 1706748 86804 ? � � � Ssl �Apr17 116:20 /usr/bin/myt
> �5.3 �2.5 1706748 95612 ? � � � Ssl �Apr17 198:05 /usr/bin/myt
> �5.4 �2.6 1706748 102480 ? � � �Ssl �Apr17 278:57 /usr/bin/myt
> �5.4 �2.8 1706748 106516 ? � � �Ssl �Apr17 358:35 /usr/bin/myt
>
> Indeed slightly weird but the trend is a slower RSS rise than before!!
>
> Udo
>

Congratulations Udo! Your long fight seems to be over (I remember some
of your first postings). I'm impressed that you never gave up. Enjoy!
And thanks to both you and the developers that make MythTV better.
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Apr 23, 2010, 5:48 AM

Post #119 of 155 (2010 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-04-22 19:54, Johan Heikkil� wrote:
> 2010/4/22 Udo van den Heuvel <udovdh [at] xs4all>:
>> On 2010-04-21 15:34, Udo van den Heuvel wrote:
>>> http://svn.mythtv.org/trac/ticket/8266#comment:21
>>>
>>> There's some progress!
>>
>> Some early figures:
>>
>> %CPU %MEM VSZ RSS TTY STAT START TIME C
>> 5.1 2.2 1706748 87208 ? Ssl Apr17 41:00 /usr/bin/myt
>> 5.2 2.2 1706748 86804 ? Ssl Apr17 116:20 /usr/bin/myt
>> 5.3 2.5 1706748 95612 ? Ssl Apr17 198:05 /usr/bin/myt
>> 5.4 2.6 1706748 102480 ? Ssl Apr17 278:57 /usr/bin/myt
>> 5.4 2.8 1706748 106516 ? Ssl Apr17 358:35 /usr/bin/myt
>>
>> Indeed slightly weird but the trend is a slower RSS rise than before!!
>>
>
> Congratulations Udo! Your long fight seems to be over (I remember some
> of your first postings). I'm impressed that you never gave up. Enjoy!
> And thanks to both you and the developers that make MythTV better.

Thanks for the congrats. You're welcome. And indeed, I only reported,
the devs did the actual work.

The logs start looking weird, to me at least:

%CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
5.1 2.2 1706748 87208 ? Ssl Apr17 41:00 /usr/bin/myt
5.2 2.2 1706748 86804 ? Ssl Apr17 116:20 /usr/bin/myt
5.3 2.5 1706748 95612 ? Ssl Apr17 198:05 /usr/bin/myt
5.4 2.6 1706748 102480 ? Ssl Apr17 278:57 /usr/bin/myt
5.4 2.8 1706748 106516 ? Ssl Apr17 358:35 /usr/bin/myt
5.4 2.6 1627816 100416 ? Ssl Apr17 438:03 /usr/bin/myt

Yes, both VSZ *and* RSS go down at least once. (!)
It's only a few days of logging with 0.23-fixes but it shows us
different phenomena.
It's not giving reason for concern, just wanting to understand what's
happening...

Kind regards,
Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

May 3, 2010, 5:45 AM

Post #120 of 155 (1909 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-04-23 14:48, Udo van den Heuvel wrote:
> Yes, both VSZ *and* RSS go down at least once. (!)
> It's only a few days of logging with 0.23-fixes but it shows us
> different phenomena.
> It's not giving reason for concern, just wanting to understand what's
> happening...

Some more numbers:

%CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
5.1 2.2 1706748 87208 ? Ssl Apr17 41:00 /usr/bin/mythbackend -
5.2 2.2 1706748 86804 ? Ssl Apr17 116:20 /usr/bin/mythbackend -
5.3 2.5 1706748 95612 ? Ssl Apr17 198:05 /usr/bin/mythbackend -
5.4 2.6 1706748 102480 ? Ssl Apr17 278:57 /usr/bin/mythbackend -
5.4 2.8 1706748 106516 ? Ssl Apr17 358:35 /usr/bin/mythbackend -
5.4 2.6 1627816 100416 ? Ssl Apr17 438:03 /usr/bin/mythbackend -
5.5 3.1 1701580 120648 ? Ssl Apr17 519:19 /usr/bin/mythbackend -
5.4 3.3 1701580 127676 ? Ssl Apr17 597:52 /usr/bin/mythbackend -
5.4 3.3 1701580 127992 ? Ssl Apr17 675:59 /usr/bin/mythbackend -
5.5 3.4 1701580 131480 ? Ssl Apr17 758:18 /usr/bin/mythbackend -
5.5 3.6 1701580 140484 ? Ssl Apr17 836:01 /usr/bin/mythbackend -
5.4 3.7 1701580 141072 ? Ssl Apr17 912:28 /usr/bin/mythbackend -
5.5 4.1 1712952 158476 ? Ssl Apr17 993:58 /usr/bin/mythbackend -
5.5 4.1 1712952 159348 ? Ssl Apr17 1075:46 /usr/bin/mythbackend
5.5 4.2 1712952 163004 ? Ssl Apr17 1154:39 /usr/bin/mythbackend
5.5 4.4 1737208 170064 ? Ssl Apr17 1232:01 /usr/bin/mythbackend

As you can see the VSZ stays fairly constant but there appears to be an
upward trend.
Same for RSS.

At least the change per day is at least one or maybe two orders of
magnitude smaller than a few months before, so we are in a much more
comfortable position right now.

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

May 26, 2010, 6:59 AM

Post #121 of 155 (1778 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-05-03 14:45, Udo van den Heuvel wrote:
> Some more numbers:
>
> %CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
5.1 2.2 1706748 87208 ? Ssl Apr17 41:00 /usr/bin/mythbackend -
5.2 2.2 1706748 86804 ? Ssl Apr17 116:20 /usr/bin/mythbackend -
5.3 2.5 1706748 95612 ? Ssl Apr17 198:05 /usr/bin/mythbackend -
5.4 2.6 1706748 102480 ? Ssl Apr17 278:57 /usr/bin/mythbackend -
5.4 2.8 1706748 106516 ? Ssl Apr17 358:35 /usr/bin/mythbackend -
5.4 2.6 1627816 100416 ? Ssl Apr17 438:03 /usr/bin/mythbackend -
5.5 3.1 1701580 120648 ? Ssl Apr17 519:19 /usr/bin/mythbackend -
5.4 3.3 1701580 127676 ? Ssl Apr17 597:52 /usr/bin/mythbackend -
5.4 3.3 1701580 127992 ? Ssl Apr17 675:59 /usr/bin/mythbackend -
5.5 3.4 1701580 131480 ? Ssl Apr17 758:18 /usr/bin/mythbackend -
5.5 3.6 1701580 140484 ? Ssl Apr17 836:01 /usr/bin/mythbackend -
5.4 3.7 1701580 141072 ? Ssl Apr17 912:28 /usr/bin/mythbackend -
5.5 4.1 1712952 158476 ? Ssl Apr17 993:58 /usr/bin/mythbackend -
5.5 4.1 1712952 159348 ? Ssl Apr17 1075:46 /usr/bin/mythbackend
5.5 4.2 1712952 163004 ? Ssl Apr17 1154:39 /usr/bin/mythbackend
5.5 4.4 1737208 170064 ? Ssl Apr17 1232:01 /usr/bin/mythbackend
5.5 4.5 1737208 173720 ? Ssl Apr17 1315:15 /usr/bin/mythbackend
5.5 4.7 1737208 180896 ? Ssl Apr17 1394:05 /usr/bin/mythbackend
5.5 4.7 1737208 180696 ? Ssl Apr17 1472:53 /usr/bin/mythbackend
5.5 4.7 1737208 182164 ? Ssl Apr17 1554:42 /usr/bin/mythbackend
5.5 4.9 1737208 186260 ? Ssl Apr17 1632:24 /usr/bin/mythbackend
5.5 5.0 1737208 193708 ? Ssl Apr17 1708:08 /usr/bin/mythbackend
5.4 5.2 1737208 198320 ? Ssl Apr17 1784:13 /usr/bin/mythbackend
5.5 5.3 1737208 203292 ? Ssl Apr17 1865:28 /usr/bin/mythbackend
5.5 5.3 1737208 204828 ? Ssl Apr17 1945:37 /usr/bin/mythbackend
5.5 5.6 1737208 213396 ? Ssl Apr17 2025:39 /usr/bin/mythbackend
5.5 5.7 1737208 219176 ? Ssl Apr17 2103:05 /usr/bin/mythbackend
5.5 5.7 1737208 220036 ? Ssl Apr17 2182:48 /usr/bin/mythbackend
5.4 5.9 1737208 226384 ? Ssl Apr17 2260:17 /usr/bin/mythbackend
5.4 6.0 1737208 230464 ? Ssl Apr17 2338:07 /usr/bin/mythbackend
5.4 6.2 1737208 236404 ? Ssl Apr17 2418:39 /usr/bin/mythbackend
5.4 6.2 1737208 239068 ? Ssl Apr17 2496:53 /usr/bin/mythbackend
5.4 6.3 1737208 242044 ? Ssl Apr17 2576:31 /usr/bin/mythbackend
5.5 6.4 1737208 244420 ? Ssl Apr17 2658:25 /usr/bin/mythbackend
5.5 6.6 1737208 254336 ? Ssl Apr17 2742:15 /usr/bin/mythbackend
5.5 6.7 1737208 257624 ? Ssl Apr17 2817:39 /usr/bin/mythbackend
5.4 6.8 1737208 260644 ? Ssl Apr17 2893:24 /usr/bin/mythbackend
5.4 6.9 1737208 264344 ? Ssl Apr17 2972:55 /usr/bin/mythbackend
5.4 7.0 1802744 267596 ? Ssl Apr17 3053:03 /usr/bin/mythbackend

As you can see VSZ grows, yet slowly. RSS also grows the past few days.

Now that 0.23-fixes is official I will do a fresh build soon and see
what that gives as I read about some (small?) leak being fixed.

Kind regards,
Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Jul 29, 2010, 7:24 AM

Post #122 of 155 (1569 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-05-26 15:59, Udo van den Heuvel wrote:
> On 2010-05-03 14:45, Udo van den Heuvel wrote:
>> Some more numbers:

Some fresh ones for 0.23-fixes:

%CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
5.2 4.3 1707288 165940 ? Ssl Jun20 1488:33 /usr/bin/mythbackend
5.2 4.4 1715484 169972 ? Ssl Jun20 1559:42 /usr/bin/mythbackend
5.1 4.3 1690896 165116 ? Ssl Jun20 1630:29 /usr/bin/mythbackend
5.2 4.7 1715484 180916 ? Ssl Jun20 1706:42 /usr/bin/mythbackend
5.2 4.8 1715484 184132 ? Ssl Jun20 1781:47 /usr/bin/mythbackend
5.1 5.0 1715484 192024 ? Ssl Jun20 1856:07 /usr/bin/mythbackend
5.1 4.9 1715484 189856 ? Ssl Jun20 1930:59 /usr/bin/mythbackend
5.1 5.1 1715484 197204 ? Ssl Jun20 2005:57 /usr/bin/mythbackend
5.2 5.3 1715484 201604 ? Ssl Jun20 2082:10 /usr/bin/mythbackend
5.2 5.3 1707288 203096 ? Ssl Jun20 2156:40 /usr/bin/mythbackend
5.2 5.4 1707288 207204 ? Ssl Jun20 2235:20 /usr/bin/mythbackend
5.2 5.5 1715484 209328 ? Ssl Jun20 2310:53 /usr/bin/mythbackend
5.2 5.7 1715484 218468 ? Ssl Jun20 2385:10 /usr/bin/mythbackend
5.2 5.8 1715484 223480 ? Ssl Jun20 2457:55 /usr/bin/mythbackend
5.2 5.8 1715484 221868 ? Ssl Jun20 2534:04 /usr/bin/mythbackend
4.5 2.0 1688160 79252 ? Ssl Jul24 25:19 /usr/bin/mythbackend -
4.8 2.1 1696356 83156 ? Ssl Jul24 97:42 /usr/bin/mythbackend -
5.1 2.2 1696496 86780 ? Ssl Jul24 175:22 /usr/bin/mythbackend -
5.1 2.5 1696496 95364 ? Ssl Jul24 253:25 /usr/bin/mythbackend -
5.2 2.6 1696496 101092 ? Ssl Jul24 330:45 /usr/bin/mythbackend -
[root [at] recorde ~]#

As you can see the backend (not the OS) was restarted while I was away.
You'll see both VSZ and RSS grow.

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

udovdh at xs4all

Nov 22, 2010, 6:48 AM

Post #123 of 155 (1210 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 2010-07-29 16:24, Udo van den Heuvel wrote:
> On 2010-05-26 15:59, Udo van den Heuvel wrote:
>> On 2010-05-03 14:45, Udo van den Heuvel wrote:
>>> Some more numbers:
>
> Some fresh ones for 0.23-fixes:

More fresh ones for 0.23-fixes:

%CPU %MEM VSZ RSS TTY STAT START TIME COMMAND
4.0 2.3 1679880 90840 ? Ssl Oct06 31:39 /usr/bin/mythbackend -
5.6 2.6 1680012 99796 ? Ssl Oct06 125:10 /usr/bin/mythbackend -
5.9 2.6 1680012 102140 ? Ssl Oct06 216:33 /usr/bin/mythbackend -
5.9 2.8 1680012 108556 ? Ssl Oct06 305:57 /usr/bin/mythbackend -
6.0 2.9 1680012 110416 ? Ssl Oct06 393:07 /usr/bin/mythbackend -
6.0 2.9 1680012 113200 ? Ssl Oct06 484:33 /usr/bin/mythbackend -
6.0 3.0 1680012 117416 ? Ssl Oct06 574:11 /usr/bin/mythbackend -
6.1 3.2 1680012 123292 ? Ssl Oct06 663:12 /usr/bin/mythbackend -
6.1 3.4 1680012 129736 ? Ssl Oct06 754:52 /usr/bin/mythbackend -
6.1 3.5 1680012 135468 ? Ssl Oct06 847:04 /usr/bin/mythbackend -
6.1 3.5 1680012 135360 ? Ssl Oct06 935:24 /usr/bin/mythbackend -
6.1 3.8 1680012 145592 ? Ssl Oct06 1025:32 /usr/bin/mythbackend
6.1 3.9 1680012 148680 ? Ssl Oct06 1117:17 /usr/bin/mythbackend
6.1 4.0 1680012 152764 ? Ssl Oct06 1207:08 /usr/bin/mythbackend
6.1 4.1 1680012 156928 ? Ssl Oct06 1296:06 /usr/bin/mythbackend
6.1 4.2 1680012 160504 ? Ssl Oct06 1385:58 /usr/bin/mythbackend
6.1 4.3 1680012 165268 ? Ssl Oct06 1474:56 /usr/bin/mythbackend
6.1 4.5 1680012 172912 ? Ssl Oct06 1562:17 /usr/bin/mythbackend
6.1 4.5 1680012 173192 ? Ssl Oct06 1649:09 /usr/bin/mythbackend
6.1 4.7 1680012 178860 ? Ssl Oct06 1739:12 /usr/bin/mythbackend
6.1 4.7 1680012 179556 ? Ssl Oct06 1830:06 /usr/bin/mythbackend
6.1 4.9 1680012 189812 ? Ssl Oct06 1920:33 /usr/bin/mythbackend
6.1 5.1 1680012 194036 ? Ssl Oct06 2011:58 /usr/bin/mythbackend
6.1 5.3 1680012 203588 ? Ssl Oct06 2101:03 /usr/bin/mythbackend
6.1 5.5 1630836 211120 ? Ssl Oct06 2186:45 /usr/bin/mythbackend
6.1 6.1 1696404 233720 ? Ssl Oct06 2271:27 /usr/bin/mythbackend
6.1 6.0 1696404 231568 ? Ssl Oct06 2363:18 /usr/bin/mythbackend
6.1 6.3 1696404 241712 ? Ssl Oct06 2454:03 /usr/bin/mythbackend
6.1 6.4 1704600 243760 ? Ssl Oct06 2544:24 /usr/bin/mythbackend
6.1 6.6 1704600 252040 ? Ssl Oct06 2638:53 /usr/bin/mythbackend
6.2 6.6 1704600 254072 ? Ssl Oct06 2731:02 /usr/bin/mythbackend
6.1 6.8 1704600 258292 ? Ssl Oct06 2819:26 /usr/bin/mythbackend
6.1 7.0 1704600 269152 ? Ssl Oct06 2907:39 /usr/bin/mythbackend
6.2 7.0 1704600 268368 ? Ssl Oct06 3000:13 /usr/bin/mythbackend
6.2 7.1 1704600 272412 ? Ssl Oct06 3093:15 /usr/bin/mythbackend
6.2 7.4 1704600 282684 ? Ssl Oct06 3183:43 /usr/bin/mythbackend
6.2 7.6 1704600 289180 ? Ssl Oct06 3277:48 /usr/bin/mythbackend
6.2 7.6 1704600 289264 ? Ssl Oct06 3370:17 /usr/bin/mythbackend
6.2 7.8 1770136 297140 ? Ssl Oct06 3458:46 /usr/bin/mythbackend
6.2 7.8 1770136 299320 ? Ssl Oct06 3547:31 /usr/bin/mythbackend
6.2 8.0 1770136 307248 ? Ssl Oct06 3641:44 /usr/bin/mythbackend
6.2 8.1 1770136 308964 ? Ssl Oct06 3734:24 /usr/bin/mythbackend
6.2 8.3 1770136 318268 ? Ssl Oct06 3827:14 /usr/bin/mythbackend
6.2 8.3 1770136 318092 ? Ssl Oct06 3922:20 /usr/bin/mythbackend
6.2 8.4 1770136 321776 ? Ssl Oct06 4017:25 /usr/bin/mythbackend
6.2 8.6 1745548 327516 ? Ssl Oct06 4103:34 /usr/bin/mythbackend
6.2 8.8 1770136 335660 ? Ssl Oct06 4188:48 /usr/bin/mythbackend

As you can see the backend (not the OS) was restarted while I was away.
You'll see both VSZ and RSS grow.

Udo
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

raymond at wagnerrp

Nov 22, 2010, 9:12 AM

Post #124 of 155 (1198 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On 11/22/2010 09:48, Udo van den Heuvel wrote:
> More fresh ones for 0.23-fixes:

0.24 has been released, and 0.23 is not likely to receive any further
support. Also, as indicated before, simply showing that mythbackend
continues to allocate more memory does not in any way help trace down a
possible memory leak.
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

gjhurlbu at gmail

Nov 22, 2010, 9:20 AM

Post #125 of 155 (1203 views)
Permalink

Re: mythbackend still eats memory: the current status [In reply to]

On Mon, Nov 22, 2010 at 6:48 AM, Udo van den Heuvel <udovdh [at] xs4all> wrote:
]> More fresh ones for 0.23-fixes:
>
> %CPU %MEM � �VSZ � RSS TTY � � �STAT START � TIME COMMAND
> �4.0 �2.3 1679880 90840 ? � � � Ssl �Oct06 �31:39 /usr/bin/mythbackend -
> �5.6 �2.6 1680012 99796 ? � � � Ssl �Oct06 125:10 /usr/bin/mythbackend -
> �5.9 �2.6 1680012 102140 ? � � �Ssl �Oct06 216:33 /usr/bin/mythbackend -
> �5.9 �2.8 1680012 108556 ? � � �Ssl �Oct06 305:57 /usr/bin/mythbackend -

<<SNIP>>

If you seriously think there is a memory leak in mythbackend, please
run valgrind (once again!) and find it. There are real reasons that
things can grow in memory, and they don't always indicate "eating"
memory or leaking.

Also, reporting on 0.23-fixes after 0.24 has been released... rather pointless.
_______________________________________________
mythtv-users mailing list
mythtv-users [at] mythtv
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-users

1 2 3 4 5 6 7

View All

Index | Next | Previous | View Threaded

Interested in having your list archived? Contact Gossamer Threads