Couchbase

[PCBC-159] incorrect timeout exceptions during required-durability function tests

Tue, 11 Dec 2012 19:59:32 -0600

When running the tests with the build mentioned above, I relatively regularly receive timeouts during the sync tests. Given that the host is localhost and there is no active workload (and it's on ssd), these seem to indicate a problem with the implementation of the required-durability functions.

[PCBC-158] assertion in removeOnePersist.phpt at Sync.inc:222

Tue, 11 Dec 2012 19:59:32 -0600

Had reasons to run the tests to package things up. I ran into this failed assertion.

001+ Assertion failed.. dumping backtrace..
001- PHP_COUCHBASE_OK
002+ #0 CBTestFramework->dieCommon(Expected TRUE) called at [/Users/ingenthr/src/php-ext-couchbase/tests/cbtestframework/CBTestFramework.inc:113]
003+ #1 CBTestFramework->assertTrue(, The document was mutated) called at [/Users/ingenthr/src/php-ext-couchbase/tests/Sync.inc:222]
004+ #2 Sync->testRemoveOnePersist()
005+ #3 ReflectionMethod->invoke(Sync Object ([] => cbtestframework_error_handler,[handle] => Resource id #10,[oo] => Couchbase Object ([] => Resource id #11))) called at [/Users/ingenthr/src/php-ext-couchbase/tests/cbtestframework/CBTestFramework.inc:195]
006+ #4 CBTestFramework->runSingleTest(testRemoveOnePersist) called at [/Users/ingenthr/src/php-ext-couchbase/tests/cbtestframework/cbtest-phpt-loader.inc:19]
007+ #5 couchbase_phpt_runtest(Sync, testRemoveOnePersist) called at [/Users/ingenthr/src/php-ext-couchbase/tests/phpt/Sync/removeOnePersist.php:3]
008+ [CBTestFramework] Expected TRUE

[PCBC-153] move flush function to RESTful flush

Tue, 11 Dec 2012 19:55:28 -0600

With Couchbase bucket types and memcached bucket types, the RESTful flush should always be used in Couchbase Server. Thus, the flush() function needs to be connected to the restful flush in libcouchbase:
https://github.com/couchbase/libcouchbase/blob/master/tools/cbc.cc#L947

[PCBC-150] make default view timeout 75s

Tue, 11 Dec 2012 20:00:24 -0600

Since the server side view timeout is 60s, we should raise the client side a bit higher. 75s seems like the right number.

It must also be tuneable through the constructor to the client or the .ini file or both.

[PCBC-149] persist_to should imply master when greater than zero

Tue, 11 Dec 2012 19:55:28 -0600

Currently persist_to in the observe array will consider the criteria satisfied when *any* combination of nodes say that the key is persisted. The client should ensure that at least one of those nodes is the master.

[PCBC-148] Key persistence checks wrong criterion

Tue, 11 Dec 2012 19:55:29 -0600

Derived from ~~MB-7161~~. I've got a fix for this, but just wanted to note it here.

[PCBC-147] Segmentation Fault (segfault) when querying view

Mon, 19 Nov 2012 01:51:53 -0600

When doing a query against a view, the php application segfaults. Set/get/delete commands all work fine.

Attached is an strace, a stack trace from gdb, and various other support files, including the repro script.

[PCBC-144] getVersion() returns memcached version, not couchbase version

Tue, 11 Dec 2012 19:55:29 -0600

For whatever reason, when doing getVersion() on the php ext the showed version is not the couchbase one. I assume its the memcached one, but I think this will only confuse people.

Is it possible to change this to the corresponding couchbase version?

array(2) {
  '192.168.56.101:11210' =>
  string(18) "1.4.4_600_g7ea975a"
  '192.168.56.102:11210' =>
  string(18) "1.4.4_600_g7ea975a"
}

Michael

[PCBC-143] segfault during lcb handle-level error handler

Dependency — Tue, 11 Dec 2012 19:55:29 -0600

This is related to ~~CCBC-120~~ (stack trace there). Will fix accordingly soon.

Basically the error handler checks for the 'ctx' which is not reset in-between calls.

The ideal solution should be for start_loop to accept a cookie argument and call lcb_set_cookie on it. This way there is always a valid cookie.

Otherwise the cookie remains a stale pointer and when the global error handler is called, it crashes.

[PCBC-142] Plug leaks reported by debug-mode php/zts

Tue, 11 Dec 2012 19:55:29 -0600

PHP_COUCHBASE_OK
[Fri Nov 9 14:45:38 2012] Script: '/sources/lcb2.0-repo/php/tests/phpt/Serialization/MixedSerializationErrors.php'
/sources/php-5.3.17/Zend/zend_execute_API.c(987) : Freeing 0x018EF248 (32 bytes), script=/sources/lcb2.0-repo/php/tests/phpt/Serialization/MixedSerializationErrors.php
=== Total 1 memory leaks detected ===PHP_COUCHBASE_OK
[Fri Nov 9 14:45:39 2012] Script: '/sources/lcb2.0-repo/php/tests/phpt/Serialization/SerializeJsonArrayMixed.php'
/sources/php-5.3.17/Zend/zend_execute_API.c(987) : Freeing 0x018086E0 (32 bytes), script=/sources/lcb2.0-repo/php/tests/phpt/Serialization/SerializeJsonArrayMixed.php
=== Total 1 memory leaks detected ===PHP_COUCHBASE_OK
[Fri Nov 9 14:45:39 2012] Script: '/sources/lcb2.0-repo/php/tests/phpt/Serialization/SerializeJsonObjectMixed.php'
/sources/php-5.3.17/Zend/zend_execute_API.c(987) : Freeing 0x026C65F8 (32 bytes), script=/sources/lcb2.0-repo/php/tests/phpt/Serialization/SerializeJsonObjectMixed.php
=== Total 1 memory leaks detected ===mnunberg@csure:/sources/lcb2.0-repo/php$

[PCBC-141] 1.1 dp releases not working on phps which do not export the symbol php_json_encode

Fri, 10 May 2013 02:43:07 -0500

This affects all php binaries which do not have the symbol "php_json_encode". This includes EL5 and EL6 based linux distributions

The library cannot load because the symbol is not found. Disabling using this symbol also means preventing views from functioning (as well as JSON serialization).

I'd like to note that this is not our bug and not our fault. We are probably not the only extension relying on php_json* functions, and package creators (specifically redhat) should expose this symbol.

Additionally, our configure script checks to see if the PHP_JSON_* constants are defined (and if not, compiles out the relevant code from views - this would make views not work, but not prevent the library from loading).

In any event, we should find a way to work around this, as this bug has been seen quite a bit.

One possible solution would be to call the php-level json encoding function. It may incur a bit of overhead from calling into php, but the function itself is more likely to be there.

[PCBC-139] PHP client needs fastlz compression to be compatible with legacy memcached clients

Dependency — Tue, 11 Dec 2012 19:55:30 -0600

The existing memcached clients for PHP (pecl-memcached and memcache) both default to using fastlz compression for values >= 2000 bytes.

The couchbase client library doesn't have fastlz compression included. It doesn't provide a useful error message when it runs into this problem. Instead, it misinterprets the stored value, tries to allocate a huge block of memory and fails. The error it gives (look, e.g., in the PHP-FPM web-access.log) is:

[01-Nov-2012 20:45:49 UTC] PHP Fatal error: Allowed memory size of 134217728 bytes exhausted (tried to allocate 8460104323416721361 bytes) in /usr/share/nginx/html/cb.php on line 32

To reproduce, just set a value >= 2000 bytes with pecl-memcached client, and try to read it with Couchbase client. Here's a sample:

<!DOCTYPE html>
<html>
<head>
<meta charset="utf-8">
<title>Couchbase Server PHP Test</title>
</head>
<body>

<p>
<?php

$host = "10.4.2.15";
$user = "Administrator";
$password = "password";
$bucketName = "default";
$persistConnection = false;
$cb = new Couchbase($host . ":8091",
        $user, $password, $bucketName, $persistConnection);

$mc = new Memcached('Pool-1');
if (count($mc->getServerList()) == 0) {
    // No servers, this pool needs to be configured
    $mc->addServer($host, 11211);
    //$mc->addServer("other host", 11211);
}

// Change this to 1999 and it will work
$valLength = 2000;
date_default_timezone_set('UTC');
$val = date(DateTime::RFC2822) . ' ';
$val .= str_repeat('X', $valLength - strlen($val));

$mc->set("a", $val);
echo("The value for 'a' is: ");
var_dump($cb->get("a"));

?>
</p>
</body>
</html>

The correct behavior is for the Couchbase client to ship, out of the box, compatible with existing Memcached client implementations. It should include fastlz support directly, or else clearly document how to add such support without requiring any unusual compilations or other hoops. The fastlz code is very small and portable, and hasn't changed since 2007, so it should present a minimal maintenance issue.

In addition, the PHP client documentation should clearly indicate what changes are needed in order for the client to be compatible with the other Couchbase clients (Java, .NET, etc.). Again, without requiring recompilation or extra hoops to jump through.

[PCBC-137] view querying needs to be more straightforward; e.g. character strings of decimal digits are serialised over the REST API as integers in view requests

Tue, 11 Dec 2012 19:59:32 -0600

Given this code:

$key = "1";
$result = $cb->view("design", "view", array("key"=>$key));

The PHP library will serialize this to a REST request as follows:

GET /bucket/_design/design/_view/view?key=1

This will be deserialized by couchbase as the integer value 1, not as the string value "1". Consequently, any key that is a string of decimal digits cannot be retrieved from a view with the PHP library.

This workaround can be used:

$key = "1";
$result = $cb->view("design", "view", array("key"=>'"'.$key.'"'));

The library will serialize this to a REST request as follows:

GET /bucket/_design/design/_view/view?key="1"

And couchbase will deserialize this as the string value "1".

[PCBC-135] Wrong parameter description for delete()

Tue, 11 Dec 2012 19:59:32 -0600

The docs API for the delete method says: ( http://www.couchbase.com/docs/couchbase-sdk-php-1.1/api-reference-update.html#table-couchbase-sdk_php_delete )

$object->delete($key [, $expiry ])

But the expire should actually be cas, as in the source https://github.com/couchbase/php-ext-couchbase/blob/master/couchbase.c#L401

[PCBC-134] Provide more information on connection failure

Duplicate — Mon, 11 Feb 2013 14:09:04 -0600

When you connect via the new Couchbase(...) is not a direct way to check for whether the connection failed.

It is not possible to check getResultCode()/getResultMessage() since this provides a unintilized couchbase

[PCBC-132] Document testing procedures

Tue, 11 Dec 2012 19:55:29 -0600

Document testing procedures

[PCBC-119] Expose timeout setting

Tue, 11 Dec 2012 19:55:29 -0600

libcouchbase has a timeout setting for connection responsiveness. This should be exposed in the php library.

The basic idea is that at libcouchbase initialization time, the PHP extension should set up the per-operation timeout to something the user specifies.

By the way, while in there, see if there are other tuneables that should be exposed.

[PCBC-108] version() returns the library version, not server version

Tue, 18 Sep 2012 10:52:13 -0500

Documentation wrongly states that version returns the version of the server; this in fact returns the version of the library

[PCBC-104] PHP Client Need to Catch and Process "Object Too Large"errors from libcouchbase

Tue, 4 Dec 2012 19:01:17 -0600

Problem Summary:

When writing more than 1MB of data using php libcouch client, the client itself does not process the object too large error and the libcouchbase library logs a PHP warning as "PHP Warning: Couchbase::set(): Failed to store a value to server: Object too big in.....".

As result, it's impossible to write a strict warning free high performance php app using the current setup when we can't control what data is attempted to be cached.

1) We cannot catch this error before passing data in since more >1MB data might be compressible to less than <1MB so we have to rely on the PHP Client to do this for us.
2) We cannot catch this on the client site since it's not an exception but a PHP error/warning.
3) We can use set_error_handler() in php to catch this but it is extremely slow and negates the performance advantage of having libcouchbase in the first place.

Recommendation:

1) Return FALSE from PHP Client if post-compressed data is more than 1MB.

/* INSERT CODE test for payload_len size here before calling line below, if more, than return FALSE */

if (!(payload = php_couchbase_zval_to_payload(value, &payload_len, &flags, couchbase_res->serializer, couchbase_res->compressor TSRMLS_CC))) {
RETURN_FALSE;
}

Notes: pecl-memcached using libmemcached lib does not throw PHP warnings when object size is over limit. Error is logged internally and is up to the programmer to read the error post-fail to determine what happened.

[PCBC-100] timeouts need to be tunable and consistent with other SDKs

Mon, 26 Nov 2012 16:04:54 -0600

Provides the consistent level of time outs with other SDKs. Provide ability to set timeouts at: 1) connection, 2) authentication, and 3) ops level.

[PCBC-99] Provide Error Handling Consistent with Other SDKs

Tue, 11 Dec 2012 16:20:20 -0600

Need to provides consistent error messages and error handling. Needs to be more consistent with other SDKS, and underlying memcached protocol

[PCBC-95] new() function docs is missing the $bucket parameter:

Duplicate — Tue, 11 Dec 2012 19:59:32 -0600

These docs are incorrect: http://www.couchbase.com/docs/couchbase-sdk-php-1.0/api-reference-summary.html

It says: "$object->new Couchbase($url, $username, $password [, $persistent ])"

It should say

[PCBC-92] Completely document constants and result codes.

Thu, 6 Dec 2012 18:37:22 -0600

There is no documentation in the PHP Couchbase Client for the constants used as options (e.g., COUCHBASE_OPT_COMPRESSOR, Couchbase::GET_PRESERVE_ORDER) and result codes (e.g., COUCHBASE_KEY_EEXISTS). This documentation is critical for usability of the library.

Also, it would be very nice to have IDE support for these things, via a PHP file that could require_once() into the app so the IDE will find these symbols, shipped as part of the package.

[PCBC-89] Update the getting started to match the Beer sample DB

Tue, 11 Dec 2012 20:06:57 -0600

Once the sample DB has been integrated into the server, update the getting started (both on the web page and in the documentation) to use the sample database.

[PCBC-88] Add a unit/integration test validation of OBSERVE + view stale=false

Tue, 11 Dec 2012 19:55:29 -0600

Ensure that when a synchronous set is used with the new low-level observe, the index is fully updated when querying with view=false.

[PCBC-87] View error options at query time

Tue, 11 Dec 2012 19:55:28 -0600

At view query time, we need to ensure the developer has the ability to add an "on_error" parameter with values of either "stop" or "continue".

[PCBC-86] Bucket management

Tue, 11 Dec 2012 19:55:29 -0600

Add the necessary features to create and remove buckets. Also, add the ability to call the RESTful bucket flush.

[PCBC-83] Highlevel synchronous mutation operations atop observe

Tue, 11 Dec 2012 19:55:29 -0600

After ~~PCBC-81~~ is implemented, the idea here is to implement synchronous mutation and deletion operations. They would, for example take the key/cas returned from the set and then poll in a loop until the desired state in the observe is reached.

There are additional examples here:
http://www.couchbase.com/wiki/display/couchbase/Observe

[PCBC-79] Tarball releases for PHP clients

Tue, 13 Nov 2012 15:08:00 -0600

It would be nice to have released tarballs to point for people having troubles with binary builds

[PCBC-73] get() method signature less optimal than it could be

Wed, 5 Dec 2012 14:34:23 -0600

It'd be great to move the $cas_token parameter to the second place as $cas_token is more likely to be used with get() than $callback. At least it would seem that way given the existence of two callback-focused/requiring functions: getDelayed and getDelayedByKey.

Due to this already being in use, it may need some type detection so that the two parameters technically become interchangeable within the signature, but we throw a deprecation warning (starting with 1.1) that states that $cas_token should come second going forward.

Additionally in 1.0.3, the callback for get() does not seem to be called--erroneous variable name output doesn't throw any errors, and basic echo's don't ever "surface" into the buffer.

Thanks.

[PCBC-72] getDelayed seems to return a Boolean not a "string (Binary object)" as the docs state

Tue, 11 Dec 2012 19:59:32 -0600

I'm not sure if this is a doc error or a code error, but the documentation for getDelayed [1] states it returns a "string (Binary object)" when in fact it returns a Boolean.

The return of a Boolean is likely the intended response as getDelayed implies the use of a callback.

I've marked this issue as being related to both docs & library, though, in case I've missed something.

[1] http://www.couchbase.com/docs/couchbase-sdk-php-1.0/api-reference-retrieve.html#table-couchbase-sdk_php_get-delayed

[PCBC-57] support errors on views

Tue, 11 Dec 2012 19:55:29 -0600

Views now have some options on capturing errors and stopping on errors as documented here:
http://www.couchbase.com/docs/couchbase-manual-2.0/couchbase-views-writing-errors.html

The client needs to support these.

[PCBC-52] add support for unlock command

Tue, 4 Dec 2012 12:12:16 -0600

Couchbase server has a keyed unlock command, which goes along with the getl command. The PHP client needs to add support for this command. Support is being added to libcouchbase.

[PCBC-51] Thoroughly unit-test API arguments

Mon, 8 Oct 2012 15:59:47 -0500

Make sure the procedural and OO interfaces work as advertise to avoid the ::increment() issue where a procedural resource was still expected.

[PCBC-47] Allow to make non-persistent connections in new-style constructor

Wed, 7 Nov 2012 15:51:09 -0600

with new Couchbase("http://server:port/bucket") you have no way of specifying that you don't want a persistent connection. You can do so with new Couchbase($server, $user, $pass, $bucket, $persistent), but it'd be nice to allow new Couchbase($url[, persistent = true]);

[PCBC-35] Add automated cluster tests

Mon, 12 Nov 2012 01:50:53 -0600

Add automated cluster tests

[PCBC-22] running integration tests under CI

Wed, 7 Nov 2012 15:52:21 -0600

running integration tests under CI

[PCBC-19] generating shippable artifacts

Tue, 11 Dec 2012 20:08:06 -0600

generating shippable artifacts

[PCBC-13] Warn for illogic view parameters

Sat, 15 Dec 2012 13:20:39 -0600

Some combinations of view parameters do not make sense e.g. group=true and group_level=N, the SDK should warn if an illogic combination is specified.

[PCBC-10] Enable SASL

Thu, 1 Mar 2012 06:19:05 -0600

we currently explicitly build without SASL bindings. We should look into enabling them, at least optionally.