[pLog-svn] r5964 - plog/branches/lifetype-1.2/class/dao
Jon Daley
plogworld at jon.limedaley.com
Tue Dec 18 18:08:44 EST 2007
It is definitely much better than it was. I thought it was
completely fixed, but I see one of my installations is still broken. I'll
have to poke around and find out where it is. It will probably be a
similar sort of bug - it is really easy to mess up the cache with those
get() and getMany() calls if the inputs aren't completely sanitized.
On Tue, 18 Dec 2007, Oscar Renalias wrote:
> So do you really think that it is finally fixed? I noticed at least
> one check-in reverting a previous one... :) Did that have anything to
> do with this?
>
> Oscar
>
> On Dec 17, 2007, at 9:02 PM, Jon Daley wrote:
>
>> woohoo! I found it. The trouble is the "new" functions for
>> setting things based on article title. We weren't checking if the
>> title
>> actually existed or not, prior to setting the cache object (and other
>> things, but less critical).
>> I am looking over the changes now - it would be great if other
>> folks, particularly those who are using custom URLs, but anyone for a
>> sanity check to make sure that the checked in code is better than
>> what you
>> have, where "better" is defined as:
>>
>> 1. Doesn't crash
>> 2. Lighter CPU load
>> 3. Doesn't have random problems
>>
>>
>> Areas that are interesting to look at:
>> Referrer counting, particularly with custom URLs,
>> and the preference that updates even when caching is turned on
>> Updating the "read" count in the same above situation
>> Searches: I originally thought the bug had to do with searches,
>> and while I am pretty sure that isn't the case, it would be
>> good to run a bunch of searches, and then view individual
>> articles, and make sure everything still views properly.
>>
>>
>>
>>
>> On Sat, 15 Dec 2007, Jon Daley wrote:
>>
>>> I have a handful of installations that are running my latest
>>> filter order change. It has been working excellently (ie. all
>>> spams have
>>> been caught by badbehavior (and/or hiddeninput) and thrown away, so
>>> the comments are not being saved to the database), BUT...
>>>
>>> I think the below bug is more visible now, probably because since
>>> there are less comments, the cache doesn't get cleaned as often,
>>> and so
>>> the bad cache values are sticking around for folks to see.
>>> The good news is that assuming I/we can find the cache bug, the
>>> cpu load going forward will be much less, since I can see the cache
>>> not
>>> being reset so often.
>>> The bad news is that I have looked for this bug before, and have
>>> not been able to find it as of yet. I'll probably need to write a
>>> cache
>>> test that runs through the various actions and see which one is
>>> putting
>>> corrupt data in the cache.
>>>
>>> On Fri, 5 Oct 2007, Jon Daley wrote:
>>>
>>>> I am happy with this change for now. More work needs to be done
>>>> to see why this happened in the first place, but this fix fixes the
>>>> problem, and shouldn't cause any downsides. I guess the only
>>>> thing is
>>>> that there might be a related bug somewhere, but I haven't seen it.
>>>>
>>>> On Sat, 6 Oct 2007, Oscar Renalias wrote:
>>>>> Well, I don't mind delaying 1.2.5 for a few more days to see if
>>>>> this
>>>>> really makes any (positive) difference... Just keep us updated.
>>>>>
>>>>> Oscar
>>>>>
>>>>> On 4 Oct 2007, at 05:24, Jon Daley wrote:
>>>>>
>>>>>> Yeah, I agree. I don't see how my fix can break anything else,
>>>>>> since it just postpones the article getting into the cache for
>>>>>> some
>>>>>> other
>>>>>> operation, but I don't really know why it fixes it either.
>>>>>> I made this change probably last Saturday or so on the two sites
>>>>>> that were having this issue, and they have been fine since then.
>>>>>> Where,
>>>>>> usually, they would have gotten a messed up cache file in one or
>>>>>> two days.
>>>>>>
>>>>>> On Wed, 3 Oct 2007, Oscar Renalias wrote:
>>>>>>> I'm not so sure... There's nothing wrong with the fix, it's
>>>>>>> just that
>>>>>>> I don't understand it :)
>>>>>>>
>>>>>>> On 10/3/07, jondaley at devel.lifetype.net
>>>>>>> <jondaley at devel.lifetype.net> wrote:
>>>>>>>> Author: jondaley
>>>>>>>> Date: 2007-10-03 14:15:27 -0400 (Wed, 03 Oct 2007)
>>>>>>>> New Revision: 5964
>>>>>>>>
>>>>>>>> Modified:
>>>>>>>> plog/branches/lifetype-1.2/class/dao/articles.class.php
>>>>>>>> Log:
>>>>>>>> This fixes the corrupt cache problem.
>>>>>>>>
>>>>>>>> I don't think this is the real fix, as I think this code was
>>>>>>>> probably okay,
>>>>>>>> but I couldn't figure out where $article was being set to an
>>>>>>>> array()
>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>> Modified: plog/branches/lifetype-1.2/class/dao/
>>>>>>>> articles.class.php
>>>>>>>> =
>>>>>>>> =
>>>>>>>> =
>>>>>>>> ================================================================
>>>>>>>> --- plog/branches/lifetype-1.2/class/dao/articles.class.php
>>>>>>>> 2007-10-02 18:55:05 UTC (rev 5963)
>>>>>>>> +++ plog/branches/lifetype-1.2/class/dao/articles.class.php
>>>>>>>> 2007-10-03 18:15:27 UTC (rev 5964)
>>>>>>>> @@ -991,7 +991,8 @@
>>>>>>>> RecentArticles::resetRecentArticlesCache
>>>>>>>> ( $article->getBlogId());
>>>>>>>> $this->_cache->removeData( $article->getBlogId(),
>>>>>>>> CACHE_ARTICLESPERMONTH );
>>>>>>>> $this->_cache->removeData( $article-
>>>>>>>>> getId
>>>>>>>> (), CACHE_ARTICLETEXT );
>>>>>>>> - $this->_cache->setData( $article->getId(),
>>>>>>>> CACHE_ARTICLES, $article );
>>>>>>>> + $this->_cache->removeData( $article->getId(),
>>>>>>>> CACHE_ARTICLES );
>>>>>>>> +// $this->_cache->setData( $article->getId(),
>>>>>>>> CACHE_ARTICLES, $article );
>>>>>>>> $this->_cache->removeData( $article-
>>>>>>>>> getPostSlug(), CACHE_ARTICLES_BYNAME );
>>>>>>>>
>>>>>>>> return true;
>>>>>>>>
>>>>>>>> _______________________________________________
>>>>>>>> pLog-svn mailing list
>>>>>>>> pLog-svn at devel.lifetype.net
>>>>>>>> http://limedaley.com/mailman/listinfo/plog-svn
>>>>>>>>
>>>>>>> _______________________________________________
>>>>>>> pLog-svn mailing list
>>>>>>> pLog-svn at devel.lifetype.net
>>>>>>> http://limedaley.com/mailman/listinfo/plog-svn
>>>>>>>
>>>>>>
>>>>>> --
>>>>>> Jon Daley
>>>>>> http://jon.limedaley.com/
>>>>>>
>>>>>> Never mess up an apology with an excuse.
>>>>>> _______________________________________________
>>>>>> pLog-svn mailing list
>>>>>> pLog-svn at devel.lifetype.net
>>>>>> http://limedaley.com/mailman/listinfo/plog-svn
>>>>>>
>>>>>
>>>>> _______________________________________________
>>>>> pLog-svn mailing list
>>>>> pLog-svn at devel.lifetype.net
>>>>> http://limedaley.com/mailman/listinfo/plog-svn
>>>>>
>>>>
>>>> --
>>>> Jon Daley
>>>> http://jon.limedaley.com/
>>>>
>>>> The sooner you fall behind, the more time you'll have to catch up.
>>>> _______________________________________________
>>>> pLog-svn mailing list
>>>> pLog-svn at devel.lifetype.net
>>>> http://limedaley.com/mailman/listinfo/plog-svn
>>>>
>>>
>>> --
>>> Jon Daley
>>> http://jon.limedaley.com/
>>>
>>> If you perceive that there are four possible ways in which
>>> a procedure can go wrong and circumvent these, then a fifth way,
>>> unprepared for, will promptly develop.
>>> -- Murphy's Sixth Law
>>> _______________________________________________
>>> pLog-svn mailing list
>>> pLog-svn at devel.lifetype.net
>>> http://limedaley.com/mailman/listinfo/plog-svn
>>>
>>
>> --
>> Jon Daley
>> http://jon.limedaley.com/
>>
>> I haven't lost my mind; I have a tape backup somewhere.
>> _______________________________________________
>> pLog-svn mailing list
>> pLog-svn at devel.lifetype.net
>> http://limedaley.com/mailman/listinfo/plog-svn
>
> _______________________________________________
> pLog-svn mailing list
> pLog-svn at devel.lifetype.net
> http://limedaley.com/mailman/listinfo/plog-svn
>
--
Jon Daley
http://jon.limedaley.com/
Whining is anger through a small opening.
-- Stuart Smalley
More information about the pLog-svn
mailing list