On Wed, 2002-12-11 at 21:04, Adon Irani wrote:> > I would check the system logs (/var/log/messages) first for panic > > messages, make sure you have decent cooling for fast processors. > > i still have to check my cooling , but the /var/log/messages is clear/no > panics .panics are rare but good to check for.> > Running something like ices (or any streamer) involves constant CPU use > > so heat could be a factor. > > according to the WHOWATCH util my cpu load was marginal (<.1 ) , i wonder > if this is only for those connected users . ( i'll look into this ) .If it's the figure thats reported from uptime then it doesn't mean much, just on average the number of runnable jobs. This will mainly depend on what's running but with ices configured with 2 streams you'll have one input and 2 instance threads running but not all the time.> > Make sure you have a recent kernel for recent hardware, there are always > > some subtle bugs that need to be sorted. The newer athlons for instance > > need some tweaks for reliability as some BIOSes don't program them > > correctly (the usual problem). > > originally i had 2.4 , but i was having many problems and had read a few > reports of icecast breaking w/ the upgrade . so on this attempt , i went > w/ 2.2 .. but i'll go back to 2.4 now .2.4 is fine, I'm running on the RH8 kernel here, which is 2.4.18 plus bits but I've run other versions without issue.> > Other things to check could be the alsa drivers, check against OSS and > > any other recent alsa driver package (www.alsa-project.org). The X > > server can be eliminated by stopping the server (init 3) and just run > > ices off the console. > > X is not installed .. and for the alsa, i've used the debian packages . > but i'll find the most recent and try those for tests sakes .Not running X means that the graphics side won't be the cause (can be awkward). I would suspect heating for the moment, I had similar problems some time ago and problems such as not burning CDs correctly until a put a fan in. Another key thing to check is memory, there's a util memcheck86 (IIRC) to do this, but building a large optimised program like the kernel with -j N can be quite good as well (make N something like 4).> > If it still happens then just check what happens to the processes over > > that period of time. vmstat 1 is usually good. > > i expected it would, and i got your msg just in time . about 75 mins in > (almost exactly the length of the last run ) it crashed . but the vmstat 1 > output is uneventful .. > > final entry > 1 0 0 0 321884 154192 22228 0 0 0 0 130 92 5 0 95 > > oldest, > n-22 er so > 0 0 0 0 304608 154192 22228 0 0 0 0 124 81 19 0 81 > > (i'll have to look into what the us register is , this is the only > variable in flux 19,20,32,5,5,23,22,3,5...5 . and the jump in free > memory , only the last 3 lines are 321884 , the rest are +/- 4k off 304608 > ).us is userspace (time spent in actual programs). If the last figure is hovering around 80-90 mark then your machine is mainly idle. It looks fine from here though. karl. <p>--- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
> On Wed, 2002-12-11 at 19:50, Adon Irani wrote: > > > > hello ., > > > > i've got icecast2/ices2 (libshout/libao ) compiled/running from cvs .> This sound like some instability of the kernel/X/hardware rather than > ices/icecast. If by tty's you mean the console then they can switch> > I would check the system logs (/var/log/messages) first for panic > messages, make sure you have decent cooling for fast processors.i still have to check my cooling , but the /var/log/messages is clear/no panics .> Running something like ices (or any streamer) involves constant CPU use > so heat could be a factor.according to the WHOWATCH util my cpu load was marginal (<.1 ) , i wonder if this is only for those connected users . ( i'll look into this ) .> > Make sure you have a recent kernel for recent hardware, there are always > some subtle bugs that need to be sorted. The newer athlons for instance > need some tweaks for reliability as some BIOSes don't program them > correctly (the usual problem).originally i had 2.4 , but i was having many problems and had read a few reports of icecast breaking w/ the upgrade . so on this attempt , i went w/ 2.2 .. but i'll go back to 2.4 now .> > Other things to check could be the alsa drivers, check against OSS and > any other recent alsa driver package (www.alsa-project.org). The X > server can be eliminated by stopping the server (init 3) and just run > ices off the console.X is not installed .. and for the alsa, i've used the debian packages . but i'll find the most recent and try those for tests sakes .> > If it still happens then just check what happens to the processes over > that period of time. vmstat 1 is usually good.i expected it would, and i got your msg just in time . about 75 mins in (almost exactly the length of the last run ) it crashed . but the vmstat 1 output is uneventful .. final entry > 1 0 0 0 321884 154192 22228 0 0 0 0 130 92 5 0 95 oldest, n-22 er so > 0 0 0 0 304608 154192 22228 0 0 0 0 124 81 19 0 81 (i'll have to look into what the us register is , this is the only variable in flux 19,20,32,5,5,23,22,3,5...5 . and the jump in free memory , only the last 3 lines are 321884 , the rest are +/- 4k off 304608 ). <p>thanks for the quick response ! i've got lots to keep me from pulling my hair out (just yet .. er, again !:) a:/, --- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.
> > > > w/ 2.2 .. but i'll go back to 2.4 now . > > 2.4 is fine, I'm running on the RH8 kernel here, which is 2.4.18 plusrunning 2.4.20 now /. new alsa drivers installed ( using the debian unstable , i figured this was recent enough ..instead of breaking dpkg ).> Not running X means that the graphics side won't be the cause (can be > awkward). I would suspect heating for the moment, I had similar problems > some time ago and problems such as not burning CDs correctly until a put > a fan in. Another key thing to check is memory, there's a util > memcheck86 (IIRC) to do this, but building a large optimised program > like the kernel with -j N can be quite good as well (make N something > like 4).wow . you were right on .. i installed a new fan last night . i've been running icecast2/ices2 for half-a-day already _ more than 10x longer than my 7o min runs . and i think my er .. m, right .. my memory is fine / -j4 worked and was speEdy .. but i was unable to track down memcheck86 ; best tip i have is to look at the linux distributions ,but i haven't yet . <p>many thanks Karl . a:/, --- >8 ---- List archives: http://www.xiph.org/archives/ icecast project homepage: http://www.icecast.org/ To unsubscribe from this list, send a message to 'icecast-request@xiph.org' containing only the word 'unsubscribe' in the body. No subject is needed. Unsubscribe messages sent to the list will be ignored/filtered.