Hi:
We just started using memcache. We are accessing it from Apache using C and
libmemcached library. As we ramp up, we notice some connection errors in
the Apache error log (see below).
[Tue Jan 13 13:03:09 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:13 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:13 2009] [error] (110)Connection timed out: MEMCACHED
Failed to get cap:zone_size:1003-1: SYSTEM ERROR\n
[Tue Jan 13 13:03:14 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:15 2009] [error] (110)Connection timed out: MEMCACHED
Failed to get cap:zone_size:1003-1: SYSTEM ERROR\n
[Tue Jan 13 13:03:16 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:18 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:20 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:20 2009] [error] (110)Connection timed out: MEMCACHED
Failed to get cap:zone_size:1003-1: SYSTEM ERROR\n
[[Tue Jan 13 13:03:22 2009] [error] (107)Transport endpoint is not
connected: MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ
FAILURE\n
[Tue Jan 13 13:03:22 2009] [error] (110)Connection timed out: MEMCACHED
Failed to get cap:zone_size:1003-1: SYSTEM ERROR\n
[Tue Jan 13 13:03:24 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:26 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:26 2009] [error] (110)Connection timed out: MEMCACHED
Failed to get cap:zone_size:1003-1: SYSTEM ERROR\n
[Tue Jan 13 13:03:26 2009] [error] (110)Connection timed out: MEMCACHED
Failed to get cap:zone_size:1003-1: SYSTEM ERROR\n
[Tue Jan 13 13:03:27 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
[Tue Jan 13 13:03:31 2009] [error] (107)Transport endpoint is not connected:
MEMCACHED Failed to get cap:zone_size:1003-1: UNKNOWN READ FAILURE\n
We are currently using about 1500 clients concurrently connected.
Here is the memcat stats from "memstat"
[prodi...@es-ut2 logs]$ memstat --servers=10.20.30.55:11211
Listing 1 Server
Server: 10.20.30.55 (11211)
pid: 30178
uptime: 58667
time: 1231882740
version: 1.2.6
pointer_size: 64
rusage_user: 265.907575
rusage_system: 739.682551
curr_items: 25800
total_items: 73152
bytes: 2717547
curr_connections: 820
total_connections: 1559
connection_structures: 1018
cmd_get: 94679858
cmd_set: 73152
get_hits: 80182459
get_misses: 14497399
evictions: 0
bytes_read: 2717547
bytes_written: 2717547
limit_maxbytes: 1073741824
threads: 1
The memcached was started with the following options...
[prodi...@es-ut2 logs]$ ps -ef | grep mem
prodicon 30178 1 1 Jan12 ? 00:16:54 /usr/local/bin/memcached -d
-m 1024 -l 10.20.30.55 -p 11211
Is there a way to avoid the above errors and allow more connections? We
would like to have about 3000+ connections to the memcached.
Let me know if there is any other / better information I can provide.
~Bill Willits
