Hello,

I am starting this thread with very little information I have from the
logs.
I can look for more details, if you can tell me what other information
can help to debug this issue.

Issue Description:

We have recently migrated to Memcached, and we are doing a load test
before pushing it to production.
But the Application Server (Tomcat) and all other Java processes on
that machine are getting crashed (killed), 10-30 minutes after
starting a load test.

The load test succeeds if the Memcached is running on the same machine
where the Application Server runs.
The issue occurs when the App Server does a frequent SET and GET over
the network to the Memcached server.

Error: I don't see any specific errors in the logs. My server is
frozen and when I tried to reconnect, it gves me a 'Time Out' error
message. It becomes responsive after a while. But by that time all
Java processes including my App Server is dead!

Question:

1. Is there any place I need to close any connections after doing a
SET / GET?
I am starting all my 100 instances of memcached clients, during server
startup.
2. Do you see anything wrong with my Server Configurations?


Versions:
Memcached:memcached : 1.4.5 on Linux 64Bit
Memcached Client: Spy Memcached : memcached-2.5

Servers:
Memcached Servers: 2 (16GB & 4 GB, AuthenticAMD, 2004 cpu mhz)
Web Apache : 1 (16 GB, AuthenticAMD, 2004 cpu mhz)
App Tomcat: 1 (16 GB, AuthenticAMD, 2004 cpu mhz)

Memcached startup params : memcached -d -m 10240 -u noaccess -l [host]
-p [port]
Memcached Client instances: 100

Please let me know what additional information can help you to
identify the cause.

Thanks for your help.
Abhilash

Reply via email to