hi all
i manage a server with cf8.
my customer decided to move server to a cloud.
from that moment, i lost my sleep...
one or twice a day jRun gets the error
"JRun closed connection".
this is a little report of situation... any help is very appreciated... 🙂
== SERVER SPECS ==
virtual machine on wmware cloud
with Linux CentOS, 32 bit
4 core CPU
3 GB Ram
1,5 GB Ram dedicated for JRun
JVM use 20% of allocated memory.
80.000 pages a day
== JRUN ==
Maximum number of simultaneous Template requests: 25
Maximum JVM Heap Size (MB): 1524
i activate jRun log. this is a optimal situation:
THREADS: Listen=22 Idle=60 Running=1 Delayed/Waiting=0
Time Delayed=0 Dropped=0 Completed=62 Time To Complete=11253 Total=83
THREADS: Listen=23 Idle=86 Running=1 Delayed/Waiting=0
Time Delayed=0 Dropped=0 Completed=134 Time To Complete=30495 Total=110
THREADS: Listen=22 Idle=77 Running=1 Delayed/Waiting=0
Time Delayed=0 Dropped=0 Completed=106 Time To Complete=19309 Total=100
before crash, the situation is similar... 😞
== DB ==
db (postgres 9.1) is in local lan on other virtual server,
no particulary data to provide.
== DURING THE CRASH... ==
this is the situation just before crash cf crash (from fusionreactor)
http://i.imgur.com/VwXts.png
http://i.imgur.com/ezzuV.png
there are no reasons for this situation.
no overload on cpu, no overload on ram.
in screenshot, over the first 3/4 page in list, other pages
are normally generated in 1/4 of second.
from this reports a see a strage "time question"...
== "TIME QUESTION"... ==
on this server happen at least two strage events...
- 1 -
in fusionreactor report, such as in seefusion report, such as in cf log,
i see a very strange long long long request.
this request taken 4,398 secs, 1h and 22m to complete. (ALWAYS 1h 22m!!!).
this happens randomly, in normal situation this page are generated
in few millisecons. moreover, the end-time of this tasks is IN THE FUTURE!
time of machine is updated regularly via ntp.
can this happen because of the virtual machine?
12/19 12:20:55 Warning [jrpp-871] - Thread: jrpp-871, template:
/jackets/audio.cfm, completed in 4,398 secs
12/19 12:23:00 Warning [jrpp-1068] - Thread: jrpp-1068, template:
/articoli/index.cfm, completed in 4,398 secs
12/19 13:00:05 Warning [jrpp-1126] - Thread: jrpp-1126, template:
/articoli/index.cfm, completed in 4,398 secs
12/19 13:23:10 Warning [jrpp-1292] - Thread: jrpp-1292, template:
/home/index.cfm, completed in 4,398 secs
- 2 -
second strange event, is the frequently timeout of cfloop, cfoutout,
cffile tags... but is not "real" timeout.
when i meet up with the error, there is no waiting time that should
generate the timeout error.
evidently cf with this error, want to say me such other,
but i don't know what... 😕
you have some idea?
i add "var" scope to cfc in my code...
== APACHE PROCESS ==
i activeted a simple log for apache connections.
sometimes, just before the crash the apache load
is bit overload..
H: apache process - C: connections on 80 port.
-- optimal situation --
H:79 C:195
H:108 C:255
H:85 C:264
H:81 C:163
H:77 C:198
H:88 C:402
H:82 C:243
H:82 C:196
-- just before crash --
H:68 C:63
H:81 C:81
H:143 C:130
H:156 C:152
H:163 C:169
H:163 C:163
-- just before crash --
H:178 C:186
H:178 C:170
H:178 C:187
H:178 C:176
H:178 C:188
H:178 C:215
== CFSERVER.LOG ==
from cfserver.log i get
12/18 18:48:36 error Error in getRealPathFromConn
12/18 21:51:00 error Error in getRealPathFromConn
12/18 21:51:00 error Error in getRealPathFromConn
12/18 21:51:03 error Error in getRealPathFromConn
12/18 21:51:03 error Error in getRealPathFromConn
but i do not think this is an important matter...
===
have you got some ideas for this strage case?
links, documents... any idea is welcome!
many thanks
--
Roberto
1
Reply
AdChoices