From: "Ruini Xue" Subject: delay in seeing newly created files from nfs client Date: Tue, 30 Oct 2007 17:13:07 +0800 Message-ID: <2bba2d6a0710300213i4c000f23x256e12eb5375a995@mail.gmail.com> Mime-Version: 1.0 Content-Type: multipart/mixed; boundary="===============1296886877==" To: nfs@lists.sourceforge.net Return-path: Received: from sc8-sf-mx2-b.sourceforge.net ([10.3.1.92] helo=mail.sourceforge.net) by sc8-sf-list2-new.sourceforge.net with esmtp (Exim 4.43) id 1Imn9s-0001Eb-Q7 for nfs@lists.sourceforge.net; Tue, 30 Oct 2007 02:13:12 -0700 Received: from nf-out-0910.google.com ([64.233.182.191]) by mail.sourceforge.net with esmtp (Exim 4.44) id 1Imn9y-0000uR-0E for nfs@lists.sourceforge.net; Tue, 30 Oct 2007 02:13:18 -0700 Received: by nf-out-0910.google.com with SMTP id 4so5090630nfv for ; Tue, 30 Oct 2007 02:13:13 -0700 (PDT) List-Id: "Discussion of NFS under Linux development, interoperability, and testing." List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: nfs-bounces@lists.sourceforge.net Errors-To: nfs-bounces@lists.sourceforge.net --===============1296886877== Content-Type: multipart/alternative; boundary="----=_Part_1214_9891748.1193735592533" ------=_Part_1214_9891748.1193735592533 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline Hello, I encounter some difficulties when testing file existence over NFS. I use access("filename", F_OK) in a while-loop to test for the existence of "filename", and there are two cases, but quit different results: 1. If the file is created on another node before the testing begins, then the access() would success in a very short time (less than 0.001 sec), and quit the while-loop. Otherwise, 1. If the file is created on another node after the test begins, the while-loop would last for quite a long time (about 20~30 sec) until the access() returns ok. I think this is not so related to 'noac', and I can not figure out what makes them different. The NFS server/client are configured with default options. Is there a safety method to test the existence of a newly created file? BTW, I am not in the list, please cc to me if you reply. Thanks! Best Andrew Below is my testing code: /**************************************************** * fexist.c * gcc -o fexit fexit.c * * test cases: (before the test, make sure no file 'test' exist) * 1. on one node create a file `test', then run `fexit' on another node (returns in l sec) * 2. run `fexit' in one node at first, and then creat file 'test' on another (returns in 20~30 sec) */ #include #include #include #include #include static double gettime() { struct timeval tp; gettimeofday(&tp, NULL); return ((double)tp.tv_sec + tp.tv_usec * 1e-6); } static int file_exist(const char* path) { struct stat sbuf; return access(path, F_OK); } int main() { double s, e; s = gettime(); while (file_exist("./test") != 0); e = gettime(); printf("Wait: %f\n", e - s); return 0; } ------=_Part_1214_9891748.1193735592533 Content-Type: text/html; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Content-Disposition: inline Hello,

I encounter some difficulties when testing file existence over NFS. I use access("filename", F_OK) in a while-loop to test for the existence of "filename", and there are two cases, but quit different results:
1. If the file is created on another node before the testing begins, then the access() would success in a very short time (less than 0.001 sec), and quit the while-loop. Otherwise,
1. If the file is created on another node after the test begins, the while-loop would last for quite a long time (about 20~30 sec) until the access() returns ok.

I think this is not so related to 'noac', and I can not figure out what makes them different. The NFS server/client are configured with default options. Is there a safety method to test the existence of a newly created file?

BTW, I am not in the list, please cc to me if you reply. Thanks!

Best

Andrew

Below is my testing code:
/****************************************************
* fexist.c
* gcc -o fexit fexit.c
*
* test cases: (before the test, make sure no file 'test' exist)
* 1. on one node create a file `test', then run `fexit' on another node (returns in l sec)
* 2. run `fexit' in one node at first, and then creat file 'test' on another (returns in 20~30 sec)
*/
#include <stdio.h>
#include <unistd.h>
#include <sys/time.h>
#include <sys/types.h>
#include <sys/stat.h>

static double gettime()
{
    struct timeval tp;

    gettimeofday(&tp, NULL);

    return ((double)tp.tv_sec + tp.tv_usec * 1e-6);
}

static int file_exist(const char* path)
{
    struct stat sbuf;
    return access(path, F_OK);
}

int main()
{
    double s, e;
    s = gettime();
    while (file_exist("./test") != 0);
    e = gettime();

    printf("Wait: %f\n", e - s);

    return 0;
}

------=_Part_1214_9891748.1193735592533-- --===============1296886877== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline ------------------------------------------------------------------------- This SF.net email is sponsored by: Splunk Inc. Still grepping through log files to find problems? Stop. Now Search log events and configuration files using AJAX and a browser. Download your FREE copy of Splunk now >> http://get.splunk.com/ --===============1296886877== Content-Type: text/plain; charset="us-ascii" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit Content-Disposition: inline _______________________________________________ NFS maillist - NFS@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/nfs --===============1296886877==--