[Prism54-devel] [Bug 33] kernel bug with higher rts

[bugzilla-daemon@mcgrof.com]

http://prism54.org/cgi-bin/bugzilla/show_bug.cgi?id=33





------- Additional Comments From john-and-kathy@shaw.ca  2004-02-05 02:26 -------
re comment from ajfa

the hangs are not gone with 2004 02 03, but the mgt commit error message is 
gone.

a reversal of fortune, as it were...(grin)...

One extra item of evidence which may be a clue.  I tried the usual snarf, and 
it pauses a long time while the card neglects to work.  Meanwhile, the console 
log shows eth transmit timeouts.    As long as I do not ctrl-C the snarf 
session (a telnet session), all remains stable.  i can login and use other 
telnet sessions. But when I ctrl-C the snarf session the whole system locks.  
It is as if the interruption and termination of snarf disconnects the session 
and somehow some software locks up the entire system.  It is not a coincidence 
since I've seen it several times.  It is not reliably repeatable since usually 
snarf just fails to resolve the web site, or else hangs at the download point.

Another clue:  I discovered the "-l" switch on ping.   It pings heavily for -l 
NNN packets then drops back to 1 per second.  A "ping -c 3 -l 20" command (23 
packets) returns 85% loss (20 lost), and locks up the machine about every 2nd 
try.

 



------- You are receiving this mail because: -------
You are the assignee for the bug, or are watching the assignee.