Hi,
On Tue, 2005-12-27 at 11:04 -0400, D Teed wrote:
Hello,
In udpcast it is normal to get some "timeout not answered" when the receiving buffer is full and the bottleneck is on the compression or write to disk. It just means the sender wanted to provide more data and the receiver couldn't handle it at that second.
i know, but at one point, i'll get only TimeOuts, the HDDs where i want to mirror to are working and then the i got an error. At the moment, i couldn't get more information, because the point of the abort is sporadic.
However the other errors are more significant and cause real failure. I'd try another client machine and see if the same happens, or boot up a Linux rescue disk/CDROM of some sort and try a memory testing program to see if that part of the hardware is reliable when pushed for 30 minutes or more.
If i could help you, message me please. I am back at office on the 2nd of January and try to rebuild the szenario.
--Donald Teed
On Tue, 27 Dec 2005, Pierre CANAL wrote:
Hello,
I use udpcast for test. I make and image of /home with or without compression perfectly. But when I want to send image , if there is no compression it's Ok (some timeout notAnswered[0]...but finally it's ok) When I use lzop or gzip, I get an error sometimes on the sender, sometimes on the sender : I have a lot of Timeout notAnswered=[0} notReady=[0] nrAns=0 nrReady=0 nrPart=1 avg=8121 and so on...
Dropping client because of timeout lzop <stdin> checksum error... or invalid operand PREEMPT and crash server ! process udp-sender Stack ... Call Trace ... or Gzip stop Bad EIPvalue SIGSEGV .... And strange thing .. With no compression the speed is 80Mbps with lzop 20 Mbps with gzip 7Mbps
My CPU is an Athlon 2400+ (not so bad), I use kernel 2.6.8 on a Debian Sarge.
Any ideas ?
Thanks
Pierre
Udpcast mailing list Udpcast@udpcast.linux.lu https://lll.lgl.lu/mailman/listinfo/udpcast