Create new data bucket failed

Definitely sounds like something is wrong there. Is this just one server? Can you send the output file of "/opt/couchbase/bin/mbcollect_info " to perry@couchbase.com?

I am experiencing a similar issue:

I install CouchBase 2.0 on a Windows Server 2008 R2 instance just fine.
If I add another bucket the console reports the CouchBase server as down and the memcached.dll starts adding application errors into my event log.

The CouchBase NT service is still running, but all servers are reported as down and the web interface reports that it cannot connect to any of the instances.

If I delete the new bucket, all restores to normal.

If I remove the default bucket all is good, if I add the bucket back everything breaks and I cannot do anything but reinstall CouchBase.

Metallemon, can you please send the output of "C:\Program Files\Couchbase\Server\bin\mbcollect_info " to 'perry@couchbase.com'?

Thanks

Perry

I had the exact same problem after uninstalling membase and installing couchbase on my DEV machine.
So i did a fresh Windows 2008 r2 install in a virtual machine and installed couchbase, and again, cant create new buckets.
After some 1 hour trying all possible bucket configuration i found out that the problem is related to the buckets name.
If you create a bucket with any uppercase chars in the name it will fail.
Any bucket whose name contains only lower case chars runs perfectly.

I didnt tried names with space or any special characters or even numbers.
I had used buckets with uppercase chars names in membase without any problem.

Errors in the log after creating buckets with uppercase chars ('Test'):

*Created bucket "Test" of type: membase

*Shutting down bucket "Test" on 'ns_1@192.168.56.153' for server shutdown

*Port server memcached on node 'ns_1@192.168.56.153' exited with status 255. Restarting. Messages: Connected to mccouch: "localhost:11213"
Connected to mccouch: "localhost:11213"
Extension support isn't implemented in this version of bucket_engine

*Control connection to memcached on 'ns_1@192.168.56.153' disconnected: {{badmatch,
{error,
closed}},
[{mc_client_binary,
cmd_binary_vocal,
6},
{mc_client_binary,
stats,
4},
{ns_memcached,
handle_call,
3},
{gen_server,
handle_msg,
5},
{proc_lib,
init_p_do_apply,
3}]}

I believe the reason for this problem is that CouchDB don't allow upper case letters in database name.

http://www.couchbase.org/forums/thread/creating-new-bucket-type-couchbas...

I'm checking with our engineering team on this as we speak.

Perry

timliaw/metallemon: can you confirm that buckets with no uppercase names will work for you?

Added a bucket called test, it all fell over. Removed that bucket and added one called staff and it worked. Not sure what to make of it.

The cleanup after you delete a bucket is not done perfectly. After a bucket is deleted the folder still exist on the server.

I have also experienced all kids of strange problems if you run servers (Windows Server 2008 R2) that blocks port 5489 (Futon port for a CouchDB).

I'm not sure if any of these things are related to your problem.

Update:

Added another bucket today called tags, and it all fell over. Stoped and restarted couchbase service and came back up.

It appears that the uppercase lettering in a bucket is in fact a known issue and something we're working to resolve (http://www.couchbase.org/forums/thread/creating-new-bucket-type-couchbas...)

metallemon, can you send me a new set of logs to investigate your particular issue?

sveinhelge and Nosralla, can you confirm that all is working now if you don't use a bucket name with upper case letters?

Perry

Creating a bucket works fine as long as I don't use upper case letters.

Works well without upper case letters

I am sending those to you now.

Problem in brief:
• CouchBase is installed on 2 nodes
• If I do no stores or gets CouchBase is stable
• If I run some views or do a store, one of the CouchBase instances will a go into pending state
• Not always immediately after the operation, but always happens.
• I frequently get ‘IP address seems to have changed ‘errors, but the IP`s are static
• Firewalls are off

Rebalancing operations fail.
Could this be an issue related to something on the network blocking communication?

Have you checked and are 100 % sure that is does not exist other servers/computers that have the same IP address as any of the servers in your cluster? (The servers/computers then would "fight" each other to get the ip address and cause instability)

Metallemon

It seems that you are hitting a system limit for open file descriptors. To work around this:

1. Edit a environment variable under: Control Panel -> System -> Advanced System settings -> Advanced tab -> Environment Variables
2. Modify MEMBASE_NUM_VBUCKETS, the default should be 16, please lower it to 4.
3. Delete all buckets
4. Stop Couchbase service
5. Reboot machine - This is so that the system account picks up the new value for MEMBASE_NUM_VBUCKETS
6. Recreate 2nd bucket and verify that the vbucket number is created correctly as specified with the environment variable.

If you see:
Crashing process
initial_call {mc_daemon,init,['Argument__1']}
pid <5067.9899.87>
registered_name []
error_info
{exit,{{badmatch,{error,system_limit}},

From mbbrowselogs, specifically the system_limit error, you will need to repeat the steps and retry with a lower NUM_VBUCKETS.

Please let us know if this gets you going.