Couchbase

[PYCBC-89] Implement add/replace

Tue, 30 Apr 2013 17:08:39 -0500

Implement add/replace

[PYCBC-88] Add support for increment/decrement

Tue, 30 Apr 2013 17:08:27 -0500

Add support for increment and decrement.

[PYCBC-87] Implement append/prepend

Sat, 11 May 2013 12:43:07 -0500

Append is needed in version 0.9.

[PYCBC-85] Client Download Link doesn't work without referrer

Sat, 4 May 2013 11:05:25 -0500

There is a HTTP Referer issue with the download link

wget "http://packages.couchbase.com/clients/python/couchbase-0.8.2.tar.gz" doesn't work

wget --referer=http://www.couchbase.com/develop/python/next "http://packages.couchbase.com/clients/python/couchbase-0.8.2.tar.gz" - this works

Please fix.

[PYCBC-75] Fixed "got not my vb error" problem that was causing a "30 second timeout" from memcached

Sat, 4 May 2013 11:07:34 -0500

Fixed "got not my vb error" problem that was causing a "30 second timeout" from the memcached client.

The code that was in there just had a little bug. I fixed a broken if statement and things started working. I tested this during a rebalance, saw the errors, and also saw my client continue to function fine.

Patch is here:
https://github.com/couchbase/couchbase-python-client/pull/14/files

[PYCBC-73] WARNING:client:unknown flags on get: 2000000

Sat, 4 May 2013 11:07:24 -0500

I was seeing these warnings suddenly from 2 days ago:
WARNING:client:unknown flags on get: 2000000
I modified the python client library to log the values, and they look good, (actually are int)
One thing might be related:
I did cbbackup and cbrestore before I saw these warnings.

[PYCBC-70] shutdown in client takes over 2 seconds

Sat, 4 May 2013 10:58:16 -0500

The shutdown function in the Couchbase object seems to take over 2 seconds.

From the discussion with the bug reporter:
@ingenthr it looks like the mc_client.done() is the culprit. the bucket is being deleted every request, calling mc_client.done() which is taking 2.3 sec to finish asa @ 2:00

oh? thanks for tracing that down! I'll get an issue filed ingenthr @ 2:00
perezd [~derek@23-24-210-193-static.hfc.comcastbusiness.net] entered the room. 2:00

client.Bucket asa @ 2:01
perezd left the room (quit: Client Quit). 2:01

that's interesting ingenthr @ 2:03
I wonder what it's doing there 2:03

is it bad to not properly shutdown? asa @ 2:08
looks like there is self._dispatcher_stopped_event.wait(2) 2:08

well, I think there's a configuration thread. it may leak something? I'm not certain to be honest ingenthr @ 2:08

comment it out and suddenly it iis very fast. asa @ 2:08
line 66 in couchbaseclient.py 2:09

heh ingenthr @ 2:09
well, this has been the area we need to fix I beleive 2:09

asa @ 2:10

[PYCBC-69] requests module update required to keep up with changing API

Sat, 4 May 2013 11:05:13 -0500

In requests 0.14, the API changed slightly and thus we need to update our use of the API.

Workaround:
Change line line 282 in couchbase/rest_client.py
- def _http_request(self, api, method='GET', params='', headers=None,
+ def _http_request(self, api, method='GET', params=None, headers=None,

[PYCBC-62] get_multi() doesn't work over binary client

Dependency — Sat, 4 May 2013 11:03:47 -0500

The MemcachedClient.get_multi() method works fine over 11211 (moxi), but fails against 11210. The implementation is using the suggested n-1 getq's + noop:
http://code.google.com/p/memcached/wiki/BinaryProtocolRevamped#Get,_Get_Quietly,_Get_Key,_Get_Key_Quietly

[PYCBC-56] document thoughts on using async-core for IO oriented parallelism

Sat, 4 May 2013 11:08:30 -0500

Related to ~~PYCBC-28~~

[PYCBC-29] Setting a non-existant key from VBucketAwareCouchbaseClient throws an exception instead of returning None

Sat, 4 May 2013 11:05:44 -0500

The docs don't match the code. We need to determine which one is correct (or which route we'd prefer taking), write some tests, and make the docs and code match.

Reported via http://disq.us/78w7hy?imp=7fd22085-2821-40da-9b3f-0e5134554a91&thread=574422822&zone=email_notification&event=shortener_click
Using couchbase-python-client v0.6.0 from the MB 2.0 DP SDK downloads, the above doesn't seem to be true. I am seeing the following exception thrown (instead of None being returned) when trying to get a non-existant key from a VBucketAwareCouchbaseClient:

couchbase.couchbaseclient.MemcachedError: Memcached error #1: Not found

Also, the get function doesn't return object but a tuple, with the object is in position 2.

[PYCBC-28] performance with python client library is sub-standard

Thu, 2 May 2013 11:24:57 -0500

Current basic tests with Python show the client library is not capable of the same kind of throughput we would expect from a basic client library. It's unclear if this owing to the fact that we're not that close to the IO or there's something just wrong in the current client library.

[PYCBC-26] Client throws intermittent "Exception in thread dispatcher-thread" errors

Sat, 4 May 2013 11:04:26 -0500

While working on the changes for 0.6, I started to get these errors. They went away after some time and I haven't seen them since (though I've used the Python client minimally since the 0.6 release). However, a user reported a similar experience on the forums today. I don't have steps to recreate, but wanted to make note of this as it's no longer isolated to my setup.

http://www.couchbase.com/forums/comment/reply/1001642#comment-form

Exception in thread dispatcher-thread (most likely raised during interpreter shutdown):
Traceback (most recent call last):
  File "/usr/lib64/python2.6/threading.py", line 532, in __bootstrap_inner
  File "/usr/lib64/python2.6/threading.py", line 484, in run
  File "/usr/lib/python2.6/site-packages/couchbase/couchbaseclient.py", line 661, in _start_dispatcher
  File "/usr/lib/python2.6/site-packages/couchbase/couchbaseclient.py", line 968, in dispatch
  File "/usr/lib64/python2.6/Queue.py", line 174, in get
<type 'exceptions.TypeError'>: 'NoneType' object is not callable

[PYCBC-24] Rebalance results in clients hanging

Sat, 4 May 2013 11:04:12 -0500

Using the latest version from Github. If you rebalance the cluster while activity is going on from a client. The client hangs with a Not My vBucket error. Even after the rebalance has finished it continues to hang.

[PYCBC-11] a disconnected or misbehaving connection may cause instability in the client library

Sat, 4 May 2013 11:06:24 -0500

This bug has not actually been observed, but it is expected that something like a stale connection (one where the server is gone, but no TCP timeout or RST packet has been received) could cause the client to continue to write to something not there forever. The client library should be more resiliant.

[PYCBC-6] some bugs on sending data

Sat, 4 May 2013 11:04:48 -0500

At times, sending data can get hung up. This is well known by Farshid, who can offer details. Just filing the bug for tracking.