Bug 761908 (GLUSTER-176) - [ glusterfs 2.0.6rc1 ] - "client SEG fault"
Summary: [ glusterfs 2.0.6rc1 ] - "client SEG fault"
Keywords:
Status: CLOSED WONTFIX
Alias: GLUSTER-176
Product: GlusterFS
Classification: Community
Component: stripe
Version: 2.0.5
Hardware: All
OS: Linux
low
medium
Target Milestone: ---
Assignee: Amar Tumballi
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2009-07-30 09:34 UTC by Vinayak Hegde
Modified: 2013-12-19 00:03 UTC (History)
4 users (show)

Fixed In Version:
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed:
Regression: RTNR
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:


Attachments (Terms of Use)

Description Vinayak Hegde 2009-07-30 06:51:28 UTC
Core produced and log files for another test case for this configuration
is also moved to /share/tickets directory.

Comment 1 Vinayak Hegde 2009-07-30 09:34:03 UTC
3 servers with stripe volume files

Single client with write-behind
and io-cache translators

running tools were:
dbench,iozone,kernel_compile,annihilate,exnihilate

core generated:

----- Frame 1 -----
#1  0x0000003756e31bf0 in abort () from /lib64/libc.so.6
----- list -----
11      #include "fuse_opt.h"
12      #include "fuse_i.h"
13      #include "fuse_misc.h"
14      #include "fuse_common_compat.h"
15      #include "fuse_lowlevel_compat.h"
16
17      #include <stdio.h>
18      #include <stdlib.h>
19      #include <stddef.h>
20      #include <string.h>



----- Frame 2 -----
#2  0x00002acc1566d277 in stripe_readv (frame=0xe68c690, this=0xdee9140, fd=0xe63dbc0, size=131072, offset=0) at ../../../../../xlators/cluster/stripe/src/stripe.c:2740
2740            ERR_ABORT (local);
----- list -----
2735            rounded_start = floor (offset, stripe_size);
2736            rounded_end = roof (offset+size, stripe_size);
2737            num_stripe = (rounded_end - rounded_start) / stripe_size;
2738
2739            local = CALLOC (1, sizeof (stripe_local_t));
2740            ERR_ABORT (local);
2741            local->wind_count = num_stripe;
2742            frame->local = local;
2743
2744            /* This is where all the vectors should be copied. */
----- print frame -----
$1 = (call_frame_t *) 0xe68c690
----- print *frame -----
$2 = {root = 0xe9e2340, parent = 0xe9dc5e0, next = 0xe9dc5e0, prev = 0xe9e2380, local = 0x0, this = 0xdee9140, ret = 0x2acc15876dde <wb_readv_cbk>, ref_count = 0, lock = 1, cookie = 0xe68c690}
----- print this -----
$3 = (xlator_t *) 0xdee9140
----- print *this -----
$4 = {name = 0xdee9120 "stripe", type = 0xdee9260 "cluster/stripe", next = 0xdee8d80, prev = 0xdee9b40, parents = 0xdeea3e0, children = 0xdee9a40, options = 0xdee91e0, fops = 0x2acc15870460, 
  mops = 0x2acc15870740, cbks = 0x2acc15870760, volume_options = {next = 0xdee92d0, prev = 0xdee92d0}, fini = 0x2acc1566ed83 <fini>, init = 0x2acc1566e9e1 <init>, notify = 0x2acc1566e413 <notify>, 
  ctx = 0xdee2010, itable = 0x0, ready = 1 '\001', init_succeeded = 1 '\001', private = 0xdeef8d0}
----- print fd -----
$5 = (fd_t *) 0xe63dbc0
----- print *fd -----
$6 = {pid = 25773, flags = 32768, refcount = 4, inode_list = {next = 0x2aaab07e1140, prev = 0x2aaab07e1140}, inode = 0x2aaab07e1100, ctx = 0xe67f7c0, lock = 1, _ctx = 0xe9e11c0}
----- print size -----
$7 = 131072
----- print offset -----
$8 = 0

Comment 2 Amar Tumballi 2009-07-30 21:37:45 UTC
The process crashed because there is memory leak (or there is no memory available). The 'cluster/stripe' translator is cleaned up, and enhanced in master branch. But in release branch, we didn't make the changes because it was very disruptive. 

Marking it as won't fix.


Note You need to log in before you can comment on or make changes to this bug.