28 replies to this topic

[greatred]

Dear Devs,

 

I`m not sure if anybody noticed this, but on my vuuno the enigma2 process started to hog at ~50% usage all the time.

The behavior is not present in the image from 2015/05/17 10:38. The following releases are affected by this problem.

Any idea which of the commits could cause enigma2 process to use more resources ?

[Erik Slagter]

It's totally normal that (any) enigma version uses over 50% cpu. It has always been that way.

[MiLo]

50% sounds suspiciously like a single thread spinning away.

Normal *CPU* usage of enigma2 is just a few percent, even when recording etc. Judging from the (lack of) responses here, it looks like you're alone in this though.

[blzr]

here on et9500 'normal' was always 0-5% of CPU...
 
and with d743e974b95b18c303dfebc27d253f6b323f08d9 reverted everything's back to normal:

CPU:   1% usr   0% sys   0% nic  97% idle   0% io   0% irq   0% sirq
Load average: 0.02 0.17 0.58 2/84 4238
  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND
 2809  2805 root     S     109m  38%   1% /usr/bin/enigma2

[littlesat]

So should we "again" recert this stuff?

[greatred]

It's totally normal that (any) enigma version uses over 50% cpu. It has always been that way.

 

Erik, are you sure 50% is normal ? Like I said I`m back to image from 17.05 and the extensive usage is gone. Enigma2 process CPU usage is between 0-5%

 

50% sounds suspiciously like a single thread spinning away.

Normal *CPU* usage of enigma2 is just a few percent, even when recording etc. Judging from the (lack of) responses here, it looks like you're alone in this though.

 

According to top, it was always one of the threads spending ~50% in user cpu time.

 

 

here on et9500 'normal' was always 0-5% of CPU...
 
and with d743e974b95b18c303dfebc27d253f6b323f08d9 reverted everything's back to normal:

CPU:   1% usr   0% sys   0% nic  97% idle   0% io   0% irq   0% sirq
Load average: 0.02 0.17 0.58 2/84 4238
  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND
 2809  2805 root     S     109m  38%   1% /usr/bin/enigma2

 

is ET9500 affected as-well ?

[blzr]

it is,

here's how it looks with up-to-date enigma:

CPU:  18% usr  30% sys   0% nic  46% idle   0% io   0% irq   3% sirq
Load average: 0.60 0.20 0.24 2/86 17788
  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND
17644 17640 root     R     109m  38%  50% /usr/bin/enigma2

 

[Pr2]

Hi,

 

On my ET9000 not updated, in info screen enigma2 date is: 2015-05-16

  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND
11200 11189 root     S     184m  65%   1% /usr/bin/enigma2

On my ET9000 updated today, in info screen enigma2 date is: 2015-05-19

  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND
  454   450 root     R     110m  39%  53% /usr/bin/enigma2

So indeed there is a problem.

 

Pr2

[MiLo]

My guess would be that the socket gets triggered (e.g. incoming connection) but bails out halfway through handing that (e.g. did not reach "accept") which results in endlessly looping through the same routine.

Just adding a few eDebugs to the code will quicky reveal what it's doing.

[ims]

I have all updated and on all boxes in standby 0-1%. I will try wakeup it...

Edited by ims, 21 May 2015 - 10:23.

[ims]

wake-up boxes:

XP1000:  80-90%

F1: 48-50%

DM800SE: cca 55%

[Jan Gruuthuse]

xp1000: éénHD 1920x1080i op 1920x1080i monitor kijken + oscam, telenet session running top.

 

CPU:  26% usr  56% sys   6% nic   0% idle   0% io   0% irq  10% sirq
Load average: 1.95 1.28 0.55 3/75 1149
  PID  PPID USER     STAT   VSZ %VSZ %CPU COMMAND
  409   404 root     R    93864  40%  90% /usr/bin/enigma2

 

90% fluctueert van 87% naar 90%

[MiLo]

blzr already pointed out that the problem is for all users, all machines.

Unless you've actually built your own E2 with extra debugging that points out what's happening, which is something that would really really help, there's no need to post any "me too" messages here.

[Erik Slagter]

Strace wilk probably shed a lot of light on this issue.

[littlesat]

shouldn't we first re-revert the specific commit?

[athoik]

Well if you like to update on a daily basis you know that your box might break.

That's why update my main box once a month or when I am sure that there are no major issues seen of forum.

Regarding the "re-revert" I suggest first to gather logs and identify the problem and why not maybe someome provide a patch that fixes the issue in order not to "re-re-revert" the commit

Edited by athoik, 21 May 2015 - 19:06.

[littlesat]

Or do both... first revert and in between find the issue.. and once found re-commit it with the fix.

[MiLo]

Think I found the issue now.

[littlesat]

few!

 

But I reverted the other stuff...

Edited by littlesat, 22 May 2015 - 19:30.

[athoik]

So the error could happen also before the getaddrinfo changes? Or enabling the getaddrinfo trigger that endless loop?