[Bug 225535] Delays in TCP connection over Gigabit Ethernet connections; Regression from 6.3-RELEASE

[bugzilla-noreply at freebsd.org]

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=225535

--- Comment #18 from Aleksander Derevianko <aeder at list.ru> ---
(In reply to Eugene Grosbein from comment #16)
No, it doesn't write anything to console. Output from the test itself is
redirected to file on disk (SSD 100Gb INTEL SSDSA2BZ100G3 6PB10362). Results
was checked from remote ssh session.

(In reply to Eugene Grosbein from comment #17)
I have switched MSI-X support off in one of the previous attempts to solve this
issue. Actually, it doesn't change anything at all in my test case.

-------------------
OK, current results from still-running test:

Computer A1:
root at fspa1:~/clock/new_res # grep times periodic.txt | awk '{print $3 " " $4 "
" $6 " " $8 " " $10;}' | sort | uniq -c
212603 send_sync 0 0 0 0
6437 send_sync 0 0 0 1
  10 send_sync 0 0 0 229
4177 send_sync 0 1 0 0
   1 send_sync 0 1 0 1


Computer A2:
root at fspa2:~/clock/new_res # !459
grep times periodic.txt | awk '{print $3 " " $4 " " $6 " " $8 " " $10 ;}' |
sort | uniq -c
122102 send_sync 0 0 0 0
100857 send_sync 0 0 0 1
  47 send_sync 0 0 0 2
  12 send_sync 0 0 0 230
   2 send_sync 0 0 0 231
 170 send_sync 0 1 0 0


I have set kern.timecounter.fast_gettime=0 yesteday, on both A1 and A2, and for
last 12 hours 6 more cases of long delays appear on A1, and 8 on A2 computer.

I will try to repeat this test with 
console="comconsole,vidconsole"
removed both on A1 and A2 computers.

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