Not able to telnet on 11211 after installing membase from source tar ball
Hi,
I have installed membase 1.7.1.1 from source tar ball. I followed the steps as mentioned in the link http://www.couchbase.org/wiki/display/membase/Building+Membase+on+Linux+... The installation goes through fine. Then I start the membase server through command ./install/bin/membase-server I get the ulimit error but the process continues and server is up.
I tried updating the /etc/security/limits.conf file but still no luck on getting away with this message.
Further to this if I try telnet to memcahced on 11211 I get a connection refused message. Tried netstat and found that memcached is listening to 11210. Telneting to 11210 established a connection, but as soon as I issue
set myvar 0 0 5
value
get myvar
The connection is closed with the message "Connection closed by foreign host". [note after set and entering a value I dont get a confirmation too saying STORED]
I have tried installing on Windows through the .exe and everything is working fine (Memcache is listening to 11211 itself).
Any idea how could I solve this. Also why is memcache listening on 11210 instead on 11211 on linux?
Let me know if you want any details from the logs (kindly let me know where can I find those log files). I find only 1 log file under install/var/lib/membase/logs/1 and not sure if I can make out something from it, as it has many special characters along with some text.
I am on Ubuntu 10.04.3 LTS
Note: I am installing this as one of the OS users and not as root [cannot install as root].
Thanks
Praveen
Hi,
Thanks for the reply.
Yes I am able to connect to cluster's console on 8091 and allocate a path for data and specify RAM size etc. One of the interesting messages that I see in the logs is as below
Port server moxi on node 'ns_1@127.0.0.1' exited with status 1. Restarting. Messages: 2011-12-23 08:00:53: (cproxy_config.c.317) env: MOXI_SASL_PLAIN_USR (13)
2011-12-23 08:00:53: (cproxy_config.c.326) env: MOXI_SASL_PLAIN_PWD (8)
2011-12-23 08:00:55: (cproxy_config.c.365) missing conflate
When I check the server nodes link under Manage I see my server listed and it shows up and running.
Can you let me know where this log file is read from? There is quite big log generated as I am experimenting from couple of days and stumbling on the same error. The reason I ask for location of logs is to get rid of the old logs and just have new ones so that I could share over here (as it will be small).
Thanks
Praveen
It looks like perhaps libconflate didn't build. If you look in it's source directory, there should be a config.log. If you look through that, you should be able to spot why.
Hi Matt,
I checked the config.log file and found the following File missing errors and some undeclared variables.
ac_nonexistent.h: No such file or directory
minix/config.h: No such file or directory
error: '__SUNPRO_C' undeclared
error: '__ICC' undeclared
Googling around showed that this could be an error due to using gcc. I updated my gcc from 4.4.3 to 4.6.2.
With new gcc I got an error from memcached and ep-engine module due to -Werror flag. I removed the -Werror flag from Makefile and the compilation went through fine. But when I check the config.log in libconflate the same errors exist along with one extra error complaining of the option -V passed.
configure:3007: gcc -m64 -V >&5
gcc: error: unrecognized option '-V'
Could you let me know why is this happening and how could I solve this.
Thanks
Praveen
Hi,
Some more details.
I see this in the output generated from make
>>
configure: WARNING: libcurl development lib not found. On Debian this is found in libcurl4-gnutls-dev. On RHEL5/Fedora11 it's in curl-devel. On RHEL6/Fedora12 it's in libcurl-devel.
<<
and
>>
configure: WARNING: libmemcached requires at least version 1.0.6 of Embedded InnoDB
<<
I tried installing the libcurl devel tools [apt-get install libcurl4-gnutls-dev] but still no luck. The warning is still seen.
Is this anything of concern that is affecting libconflate.
Thanks
Praveen
Here is my config.log file from libconflate component
Why do you need to build from source? Just use apt or yum.... or the .deb or .rpm file depending on your distro...
Also be sure to leave the "default" bucket intact.
Hi,
I want to try plugging in other third party persistence for membase. Hence am trying to build membase from source.
Thanks
Praveen
Check around line 1991 in your gist. It looks like your libcurl was built with some kind of openldap support which it's referencing but isn't installed. I'd use ldd to verify this.
Interesting comment on looking at another backend. Which one are you considering?
Hi Matt,
Thanks, yes the openldap was the issue. This had happened because of chnage in version during upgrade of OS to 11.04 of Ubuntu. I had to link the openldap in /usr/lib to /usr/local/lib
>>
Interesting comment on looking at another backend. Which one are you considering?
<<
I am looking at DB2 as an option.
Thanks
Praveen
Hi Matt,
Any pointers to where could I start from. Some documents or links that I can refer to which talk of the overview design of ep-engine and the basic skeleton that a persistence engine will need to follow.
I went through Trond's blog on persistence for Memcached, was good to get understanding on what is happening. I am now looking at if any docs are available specific to Membase to allow plug-in other persistence. If any could you share.
Thanks
Praveen
It's normal for the memcached process to listen on 11210. It's the moxi process which listens on 11211 and it should be started by the cluster manager, which runs in a beam.smp erlang VM.
Can you connect to the cluster's console on port 8091? If so, can you spot anything in the logs about why it won't start?