Opened 13 years ago

Closed 13 years ago

#1863 closed defect (wontfix)

New memleak in newcamd when unclean shutdown and multiple logins

Reported by: Madmaxx Owned by:
Priority: minor Component: General
Severity: medium Keywords: MEMLEAK NEWCAMD MANYPORTS
Cc: Sensitive: no

Description

I found a new memleak in actual trunk with newcamd when an "unclean" login/logout occours (=DSL Zwangstrennung).

I have only two newcamd-proxy users, uniq=0, and every user loggs in 7 times with the same user/password but with 7 different ports.

In case of "Zwangstrennung" or other similar IP-interuption the actual session is not cleaned up sucessesfully. A few seconds later the client comes back and logged in twice. At this moment the clean-up process for the old sessions is not working well and can cause highload 100%, hunging thread and rise the oscam memory usage from 7,7 MB to 8,5 MB usage.

This can be one more reason for a 100% load problem that is hard to find. It looks like a timing problem, when 1 user opens up 7 ports at once with new ip-session, when the old sessions is stalling or somewhat else "halfopen" or still active on oscam ?

Maybe another problems i found in via-reader, that rises memory usage after some timg:

One client send from time to time invalid ecm requests that results in this:

2011/05/17 10:30:11 40104580 c user1 (0500&FFF430/1136/3C:475F): not found (1 ms) by srgreader (of 1 avail 1) (Unknown ECM type) <-!!!

When i receive this type of crap many times a day, memory usage is rising, so handling in via-reader is not 100% clean?

Change History (6)

comment:1 by pooyair, 13 years ago

Resolution: wontfix
Status: newclosed

sorry ,for memory leaks ,you might provide valgrind log as explained.
http://www.streamboard.tv/wbb2/thread.php?postid=381036#post381036
and upload your log there .this ticket without log (valgrind) could not be enough for fix...
thanks

comment:2 by Madmaxx, 13 years ago

Resolution: wontfix
Status: closedreopened

Please stop closing tickets, when it is described, that NOTHING useful is loggable!

So as replacement there is a best case description of the problem. Furthermore i wrote that the problems comes up after many hours. Means: Is nessasaary to watch for a longer time at developer side.

Have you self checked 1-3 days with 9 locals and newcamd-only and some received crap-ecms requests before closing the ticket ? Have you checked "Zwangstrennung" with unclean shutdown and multiple Ports ? Fine... then do it please!

When someone gave me a hint how to trigger/log some things i will make it!

comment:3 by Madmaxx, 13 years ago

By the way: valgrind on a small openwrt router ???

comment:4 by pooyair, 13 years ago

valgrind on a small openwrt router ???
if this memory leaks related to oscam , Newcamd protocol with your 9 local cards , u could run valgrind on your linux pc too, and then upload your log ...

if u want your problem about memory leaks would be fix , u have to provide min requirements .Valgrind log could much easily and better refer devs to the part of code which need some fix than your descriptions ...
let me assure u , like your other ticket (1851) about cpu high usages, nobody couldn't help u unless u provide log

Last edited 13 years ago by pooyair (previous) (diff)

comment:5 by Madmaxx, 13 years ago

For now i have only 10 MB RAM left free on xscale openwrt router and in 1-3 days is a jump to 100CPU load at unkown point. so at which time with which package and which not-have memory i can provide a "log" ?

first i need for openwrt/wrt350nv2 a working valgrind binary AND real ram to run it maybee...

comment:6 by john_28, 13 years ago

Resolution: wontfix
Status: reopenedclosed

provide log in ticket systems , or create a topic in forum and discuss your problem there ...

Note: See TracTickets for help on using tickets.