Bug 765496 (GLUSTER-3764) - [glusterfs-3.2.5qa4]: fs-perf test blocks the filesystem
Summary: [glusterfs-3.2.5qa4]: fs-perf test blocks the filesystem
Keywords:
Status: CLOSED WORKSFORME
Alias: GLUSTER-3764
Product: GlusterFS
Classification: Community
Component: replicate
Version: pre-release
Hardware: x86_64
OS: Linux
medium
high
Target Milestone: ---
Assignee: Pranith Kumar K
QA Contact:
URL:
Whiteboard:
Depends On:
Blocks:
TreeView+ depends on / blocked
 
Reported: 2011-10-31 06:57 UTC by Raghavendra Bhat
Modified: 2011-11-29 10:56 UTC (History)
2 users (show)

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


Attachments (Terms of Use)

Description Raghavendra Bhat 2011-10-31 06:57:36 UTC
Replicate volume with replica count 2. 1 fuse client.

ran fs-perf-test with number of fd's to be opened as 4096. While it was going on killed a brick , slept and nrought it up. It blocked the entire filesystem.Statedump says there are huge number of calls (around 20,000 call stacks in the statedump, once it was around 35,000) going on most of them being writes.


[global.callpool.stack.19870.frame.5]
global.callpool.stack.19870.frame.5.ref_count=0
global.callpool.stack.19870.frame.5.translator=mirror-client-0
global.callpool.stack.19870.frame.5.complete=1
global.callpool.stack.19870.frame.5.parent=mirror-replicate-0
global.callpool.stack.19870.frame.5.wind_from=afr_nonblocking_inodelk
global.callpool.stack.19870.frame.5.wind_to=priv->children[i]->fops->finodelk
global.callpool.stack.19870.frame.5.unwind_from=client3_1_finodelk_cbk
global.callpool.stack.19870.frame.5.unwind_to=afr_nonblocking_inodelk_cbk

[global.callpool.stack.19870.frame.6]
global.callpool.stack.19870.frame.6.ref_count=1
global.callpool.stack.19870.frame.6.translator=mirror-replicate-0
global.callpool.stack.19870.frame.6.complete=0
global.callpool.stack.19870.frame.6.parent=mirror-write-behind
global.callpool.stack.19870.frame.6.wind_from=wb_sync
global.callpool.stack.19870.frame.6.wind_to=FIRST_CHILD(sync_frame->this)->fops->writev
global.callpool.stack.19870.frame.6.unwind_to=wb_sync_cbk

[global.callpool.stack.19871]
global.callpool.stack.19871.uid=0
global.callpool.stack.19871.gid=0
global.callpool.stack.19871.pid=0
global.callpool.stack.19871.unique=0
global.callpool.stack.19871.type=0
global.callpool.stack.19871.cnt=1

[global.callpool.stack.19871.frame.1]
global.callpool.stack.19871.frame.1.ref_count=0
global.callpool.stack.19871.frame.1.translator=glusterfs
global.callpool.stack.19871.frame.1.complete=0

Comment 1 Pranith Kumar K 2011-11-18 00:34:50 UTC
Johnny,
     Since this bug is not reproducible, could you let me know what needs to be done for this bug?.

Pranith

Comment 2 Raghavendra Bhat 2011-11-29 07:56:46 UTC
Tried with latest release-3.2 head. It did not get reproduced. Moving it to the resolved state. Please reopen it if found again.


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