[Case 74677] EACCELERATOR can't restore parent class

[benito]

I have the same problem. No mod_ruid2 enabled.

 

[ScottTh]

Thanks everyone for the updates about this issue.

The EasyApache team is continuing to research this and is ready to to start implementing a fix in the near future.

Please keep checking this thread as well as the EasyApache Change Log for updates:

EasyApache Change Log

 

[quanin]

I ran into this problem as well, also not running mod_ruid2 but running SUPHP and SuExec. The extension lasted less than 12 hours before I killed it, because any site with any kind of PHP at all was throwing 500 errors and leaving the same messages in the error log.

 

[inthukha]

An internal case is open to address this issue. For reference, the case number is 74677. It looks like this is happening on systems with mod_ruid2 enabled. PHP and eAccelerator should still operate as expected, despite the warning message, because permissions are setup correctly when the directory is created. This thread will be updated with more information as it's available.

Hello,


I made this topic and i confirm that there is not mod_ruid2 enabled or compiled. I believe this happen due to compilation of eAccelerator under nobody, instead root.

THANKS

 

[jecro77]

Same problem here with Mod Ruid2 and EAccelerator. After update to PHP 5.3.27 I get some errors:

PHP Fatal error: Cannot access parent:: when current class scope has no parent
PHP Fatal error: Call to undefined method

Apache/2.2.25 (Unix) mod_ssl/2.2.25 OpenSSL/0.9.8e-fips-rhel5 mod_bwlimited/1.4 PHP/5.3.27

I have to uninstall EAccelerator.

 

[digitalmahdi]

Just updated my php to PHP 5.4.17 and it happened for me too!
i am using fcgi and there is no mod ruid2 enabled for me!
i did uninstall the extension using "/scripts/phpextensionmgr uninstall EAccelerator" but i still see these lines appearing in error_log.

 

[NitroHost]

Just updated my php to PHP 5.4.17 and it happened for me too!
i am using fcgi and there is no mod ruid2 enabled for me!
i did uninstall the extension using "/scripts/phpextensionmgr uninstall EAccelerator" but i still see these lines appearing in error_log.

Maybe these are stupid questions, but did you:

1) Restart Apache after you disabled it?

2) Check /usr/local/lib/php.ini to make sure the EAccelerator lines were removed?

3) If you're using CloudLinux, force a CageFS Skeleton update so that users are pulling from the updated php.ini?

 

[digitalmahdi]

Ya i was using Cloudlinux with CageFS and after i did a force update the problem solved!

 

[ikillbill]

any updates on this please?

 

[cPanelMichael]

any updates on this please?

Further testing has indicated the issue is reproducible when mod_ruid2 is disabled. It's likely a patch for EAccelerator will be incorporated to resolve this issue. However, there is currently no specific time frame available on when that will be pushed out.

Thank you.

 

[NitroHost]

any updates on this please?

If it helps, we've found that XCache is working just fine with a good configuration set up under mod_fcgid. So much so that we probably won't go back to EAccelerator when it's fixed.

 

[rasa]

i have same problem
with PHP5.3.x + Mod FCGI or even SUPHP ! but this just happen after we setup our remote database server ! !
mod_ruid is not enabled

 

[rasa]

hello
disabling this extension could cause extra load on server and its going to make problem on heavy load environment

does cpanel have any time frame to resolve this issue ?

 

[NitroHost]

hello
disabling this extension could cause extra load on server and its going to make problem on heavy load environment

does cpanel have any time frame to resolve this issue ?

Why not disable it and use XCache instead, which is working perfectly fine?

 

[ikillbill]

why cpanel is so slow on this?

and to Nitrohost, how to tweak Xache to the best?

 

[NitroHost]

and to Nitrohost, how to tweak Xache to the best?

Hard to say without knowing your setup, what PHP handler you're using, how much RAM you want to allocate to opcode caching etc.

 

[ScottTh]

why cpanel is so slow on this?

and to Nitrohost, how to tweak Xache to the best?

Hi ikillbill,

We are currently working on this solution. We hope to have a fix available very soon. I'm watching this thread and will update it when we have released the next version of EasyApache with the fix.

Thank you for your patience!

 

[rasa]

Hard to say without knowing your setup, what PHP handler you're using, how much RAM you want to allocate to opcode caching etc.

Hello
mine that have problem is a E5-1620 + 32GB RAM + Apache 2.4 + UnixyVarnish + FastCGI or SUPHP

 

[NitroHost]

This should get you close. You may or may not want to turn the optimizer on, we found it caused problems with a couple of users' scripts. Also you may want to turn admin auth on and set a username / password (there's info on that at the XCache website). It's not really needed under mod_fcgid as the end user can only see their own cache if they upload the admin tools. Actually given that under mod_fcgid it's cache-per-process, they can only see their own cache for the process that they happen to connect to.

You'll probably want to tweak the xcache.size value too.

Btw you mentioned UnixyVarnish. This is a total aside, but if this is a shared server and you care about your users having accurate stats, and you care about billing them for the bandwidth that they use, you should know that any hits on the Varnish cache are not logged at all. Found that out the hard way with a customer pulling in 2 million+ visitors per month and north of 10 million page views. He was logging close to an 80% cache hit rate and none of that considerable bandwidth usage was showing up in his account. All of his stats were off too.

extension="xcache.so"
xcache.admin.enable_auth="off"
xcache.admin.pass=""
xcache.admin.user=""
xcache.cacher="On"
xcache.coredump_directory=""
xcache.count="1"
xcache.coveragedump_directory="/tmp/pcov/"
xcache.coverager="Off"
xcache.gc_interval="0"
xcache.mmap_path="/dev/zero"
xcache.optimizer="Off"
xcache.readonly_protection="Off"
xcache.size="32M"
xcache.slots="8K"
xcache.test="Off"
xcache.ttl="0"
xcache.var_count="1"
xcache.var_gc_interval="300"
xcache.var_maxttl="0"
xcache.var_size="0"
xcache.var_slots="8K"
xcache.var_ttl="0"

Hello
mine that have problem is a E5-1620 + 32GB RAM + Apache 2.4 + UnixyVarnish + FastCGI or SUPHP

 

[ScottTh]

Hi everyone,

EasyApache 3.22.5 has been released. The change can be seen here: EasyApache Change Log

Included in the release is a fix for the permissions issue described in this thread. Stay tuned as we're preparing another eAccelerator fix to be released very soon.

Thanks to everyone for your patience while we worked on these issues.