Drupal Crashes when writting utf8 to page cache or someone else in mongodb

mongo.utf8 int
If an exception should be thrown for non-UTF8 strings. Until version 1.0.4, the PHP driver would ignore non-UTF8 strings, even though you're not supposed to insert them. As of 1.0.4, the driver throws a MongoException. To ease the transition for applications that insert non-UTF8 strings, you can turn this option off to emulate the old, non-exception-throwning behavior. This option will be eliminated and exceptions always thrown for non-UTF8 strings starting with version 1.1.0.

Which means the fix is not exactly viable in the long term.

Read on mongodb.org:

You can definitely save compressed data, just use MongoBinData instead of saving the plain string.

Subscribing

Tested under Drupal6

131,132c131,136
< if ($cache->serialized) {
< $cache->data = unserialize($cache->data);
---
> if($cache->binary && $cache->serialized) {
> $cache->data = unserialize($cache->data->bin);
> } else if(!$cache->binary && $cache->serialized){
> $cache->data = unserialize($cache->data);
> } else {
> $cache->data = $cache->data->bin;
134c138
<
---
>
145a150
> 'binary' => true,
147c152
< 'data' => $scalar ? $data : serialize($data),
---
> 'data' => $scalar ? $data : serialize($data),
149c154
<
---
> $entry['data'] = new MongoBinData($entry['data']);
151d155
<

Thanks dob_.

Here's a patch improves it a bit.

Patch looks good, I incorporated it into Cache Tags, ran the benchmarks, and it wasn't noticeably slower. Plus it fixed the crash :) I also think the try/catch wrapping is good because I like rendered pages better than crashed pages.

This might be a newbish question but why are the $headers being removed from set()?

Also, why is there a new question mark in comments on its own line?

Sorry forgot to mention that the $headers was in D6 and moved into cache data in D7. For reference, see function drupal_page_set_cache() (or function page_set_cache() for D6).

Patch seems good, removed the question mark (comment #8). Patch commited to latest 7.x-1.x.

Hi,
I got this error using latest dev:

>
MongoException: non-utf8 string:  in MongoCollection->save() (line 198 of /sites/all/modules/mongodb/mongodb_field_storage/mongodb_field_storage.module)

Is this patch may be not complete?
Greetings, Martijn

May be something like this on mongodb_field_storage.module:

// Use MongoBinData for non-UTF8 strings.
if (is_string($new_entity) && !drupal_validate_utf8($new_entity)) {
    $new_entity = new MongoBinData($new_entity);
}

Just above line 198?

I am not a programmer, but just trying to get rid of this error..
Greetings, Martijn

New title what I think is relevant, I add a new issue if you think this is not related?
Than I am sorry to open this closed issue.
greetings, Martijn

Hi,

Edit: mongo.utf8 = 0 worked, but this is not possible on a shared host always..
I made a new issue, may be they are related, but I cannot see that: http://drupal.org/node/1944188
Sorry for opening a closed issue earlier.

Greetings, Martijn