| Re: APC issues: locking up all processes and lack of PHP 5.4 support [message #180864 is a reply to message #180858] |
Fri, 22 March 2013 12:58  |
Jerry Stuckle
Messages: 2598
Registered: September 2010
Karma:
|
Senior Member |
|
|
On 3/21/2013 11:28 PM, sroussey(at)gmail(dot)com wrote:
> On Wednesday, March 13, 2013 1:11:52 PM UTC-7, Jerry Stuckle wrote:
>> On 3/13/2013 3:21 PM, sroussey(at)gmail(dot)com wrote:
>>
>>> In one of my installs of PHP, I have about 100 servers running PHP 5.3.something (whatever RedHat has as stable on that platform). Every couple of days one of the servers locks up, with all the processes stuck in a lock wait. It looks like the process that created the lock crashed, all the others get stuck and that server needs a bullet.
>>
>>>
>>
>>> I would like to move to 5.4, but it is not stable yet, due to our dependency on APC.
>>
>>>
>>
>>> Has anyone noticed this issue as well? If so, did moving to 5.4 help? For this setup we only have "stable" installs for deployments, but it is a big enough issue that we are at least considering altering that.
>>
>>>
>>
>>> In other installs of around the same size, I don't see this issue, but I install my own PHP, so it is the most current of 5.3. Again, with no stable APC, I haven't even bothered with 5.4 yet, but in that deployment it is not an issue nor needed.
>>
>>>
>>
>>> Many thanks in advance!
>>
>>>
>>
>>> -steve--
>>
>>>
>>
>>
>>
>> Steve,
>>
>>
>>
>> I haven't had any problems running 5.3.3 on Debian, but then you
>>
>> obviously have additional packages (i.e. I don't have APC running).
>>
>>
>>
>> Is there anything in your server logs? No matter what you have in yours
>>
>> scripts, the process shouldn't crash.
>>
>>
>>
>> Without knowing the cause of the crash, it's impossible to determine if
>>
>> an upgrade will be of any help.
>>
>
> It is the APC locking up that is the problem, though a crash might be the source. Tried attaching and doing a trace and all that, but couldn't gleam enough info. It is also hard since it happens on loaded servers, not less utilized ones.
>
At the risk of repeating myself - is there anything in the server logs?
How about the system logs?
Did you configure the core dump environment, and if so, did you get a dump?
Locking problems are usually (but not always) a secondary problem. We
need to get back to the primary one.
And it's hard to catch intermittent bugs like this, especially when they
only happen on a very heavily loaded system.
--
==================
Remove the "x" from my email address
Jerry Stuckle
JDS Computer Training Corp.
jstucklex(at)attglobal(dot)net
==================
|
|
|
|
Calendar
Search
Help
Members
Register
Login
Home
]