[peruser] graceful and peruser

Alexey Vlasov renton at 1gb.ru
Sat Dec 1 04:38:59 MST 2007


Hi.

I noticed that appeared more complaints on 503 errors and disconnects (I use
peruser as backend Apache). When I looked through log files, it turned out
that the time of these errors is coincide with graceful restarts peruser
Apache.
I made the following:

$ wget --limit-rate=200k http://somehost.org/somefile.zip
--12:52:41--  http://somehost.org/somefile.zip
           => `somefile.zip'
Resolving somehost.org... 81.81.26.66
Connecting to somehost.org|81.81.26.66... connected.
HTTP request sent, awaiting response... 200 OK
Length: 104,857,600 (100M) [application/zip]

56% [=============================================>
] 59,641,128   200.81K/s    ETA 03:50

The connection at server-status disappeared at once after the restart,
though it should have remained, but with G flag. Approximately 5 minutes
later after graceful the session closed.


12:57:43 (192.44 KB/s) - Connection closed at byte 59641128. Retrying.

--12:57:43--  http://somehost.org/somefile.zip
  (try: 2) => `somefile.zip'
Connecting to somehost.org|81.81.26.66... connected.
HTTP request sent, awaiting response... 206 Partial Content
Length: 104,857,600 (100M), 45,216,472 (43M) remaining [application/zip]

68% [++++++++++++++++++++++++++++++++++++++++++++++=========>
] 71,722,240   198.23K/s    ETA 02:44

And here is the log file. 
And by the way, why after file downloading by wget appear 2 procecces:

top - 13:38:48 up 52 days, 18:26,  5 users,  load average: 11.43, 9.27,
11.94
Tasks: 1167 total,   7 running, 1154 sleeping,   0 stopped,   6 zombie
Cpu(s): 18.8% us,  8.1% sy,  0.0% ni, 46.7% id, 25.2% wa,  0.4% hi,  0.9% si
Mem:   8163252k total,  8066840k used,    96412k free,   508352k buffers
Swap:  7815580k total,   242816k used,  7572764k free,  2651384k cached

  PID USER      PR  NI  VIRT  RES  SHR S %CPU %MEM    TIME+  COMMAND
 8501 w_linux_  15   0  144m  53m 1796 S    0  0.7   0:00.02 httpd
 8508 w_linux_  25   0  144m  52m  812 S    0  0.7   0:00.00 httpd

[Sat Dec 01 13:41:34 2007] [notice] Graceful restart requested, doing
restart
[Sat Dec 01 13:41:34 2007] [notice] SIGUSR1 received.  Doing graceful
restart
[Sat Dec 01 13:41:36 2007] [notice] Apache/2.2.3 (Unix) PHP/5.2.3 configured
-- resuming normal operations
[Sat Dec 01 13:41:36 2007] [notice] child pid 16658 exit signal Segmentation
fault (11)
[Sat Dec 01 13:41:36 2007] [notice] child pid 10310 exit signal Segmentation
fault (11)
[Sat Dec 01 13:41:36 2007] [notice] child pid 12312 exit signal Segmentation
fault (11)
[Sat Dec 01 13:41:36 2007] [notice] child pid 12837 exit signal Segmentation
fault (11)
[Sat Dec 01 13:41:36 2007] [notice] child pid 14128 exit signal Segmentation
fault (11)
...
[Sat Dec 01 13:42:20 2007] [warn] long lost child came home! (pid 16288)
[Sat Dec 01 13:42:23 2007] [warn] long lost child came home! (pid 16283)
[Sat Dec 01 13:42:27 2007] [warn] long lost child came home! (pid 16305)
[Sat Dec 01 13:42:29 2007] [warn] long lost child came home! (pid 16286)

And there's no mentioning of pids 8501, 8508 in log file. 

--
BRGDS. Alexey Vlasov.



More information about the Peruser mailing list