Search squid archive

Re: assertion failed: store_swapout.cc:315: "mem->swapout.sio == self

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



Well it happens that 8GB of memory will not be enough for 240GB of ufs disk cache as far as I can tell.
Lets try to fix it one step at a time.
Since you are getting into a position which squid get's stuck we can go from step A to B and then to C.

First decrease the number of ufs directories to something like 1...
The cache_dir can handle more then what some can imagine.
What is the cache_dir number of objects?
You can use the cache_mgr interface to get some statistics on the cache_dir.
It's version 3.1.20 so basically you should be able to access it using "squidclient mgr:menu"
I do not remember the exact name but it has something about the cache_dir.
After you will have one cache_dir which fits your number of files in a form of for example 64 top directories and 256 upper directories you might see something else. note that 50M (million right?)json objects are huge but you might have a limit from the INODE limit of the FS way before squid limits.

The machine basically doesn't meet the basic specs to store 50Million files.

Eliezer


On 05/04/2014 08:27 PM, viral shah wrote:
Hi Eliezer,

Thanks for your response, please find the answers below.

what are the machine specs?
     Machine on which Squid is running, is VM having 2 cores and 8 GB of memory.

why are you using so many ufs cache dirs? is  there a reason?
      Yes, our purpose is to cache more than 50M json objects.
      Tried running with single cache directory. that didn't work,
after 10-15 days Squid will stop responding with below error in
cahce.log.
     "assertion failed: filemap.cc:76: "fm->max_n_files <= (1 << 24)"
     I came across work around for the above issue on internet, that
mentioned having multiple cache directories.


Viral



On Sat, May 3, 2014 at 11:01 AM, Eliezer Croitoru <eliezer@xxxxxxxxxxxx> wrote:
Hey,

couple questions:
- what are the machine specs?
- why are you using so many ufs cache dirs? is  there a reason?

Mainly this squid version is old and I suspect it's not supported anymore.
64 doubles 5.1 is above 240GB of cache which requires lots of memory.

Eliezer


On 05/03/2014 01:39 AM, viral shah wrote:

Hello,

I am running squid as http caching server, after running for few days
our squid server will stop responding to client requests.

When this occurs, I see below errors in cache.log.







[Index of Archives]     [Linux Audio Users]     [Samba]     [Big List of Linux Books]     [Linux USB]     [Yosemite News]

  Powered by Linux