Bug 1736342 - potential deadlock while processing callbacks in gfapi
Summary: potential deadlock while processing callbacks in gfapi
Keywords:
Status: CLOSED NEXTRELEASE
Alias: None
Product: GlusterFS
Classification: Community
Component: libgfapi
Version: 5
Hardware: All
OS: All
unspecified
high
Target Milestone: ---
Assignee: bugs@gluster.org
QA Contact: bugs@gluster.org
URL:
Whiteboard:
Depends On: 1733166 1736345
Blocks: 1733520 1736341 1737313
TreeView+ depends on / blocked
 
Reported: 2019-08-01 17:01 UTC by Soumya Koduri
Modified: 2019-08-06 11:07 UTC (History)
3 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of: 1733166
Environment:
Last Closed: 2019-08-05 05:06:51 UTC
Regression: ---
Mount Type: ---
Documentation: ---
CRM:
Verified Versions:
Embargoed:


Attachments (Terms of Use)


Links
System ID Private Priority Status Summary Last Updated
Gluster.org Gerrit 23151 0 None Merged gfapi: Fix deadlock while processing upcall 2019-08-05 05:06:50 UTC

Description Soumya Koduri 2019-08-01 17:01:13 UTC
+++ This bug was initially created as a clone of Bug #1733166 +++

Description of problem:

While running parallel I/Os involving many files on nfs-ganesha mount, have hit below deadlock in the nfs-ganesha process.


epoll thread: ....glfs_cbk_upcall_data->upcall_syncop_args_init->glfs_h_poll_cache_invalidation->glfs_h_find_handle->priv_glfs_active_subvol->glfs_lock (waiting on lock)

I/O thread:

...glfs_h_stat->glfs_resolve_inode->__glfs_resolve_inode (at this point we acquired glfs_lock) -> ...->glfs_refresh_inode_safe->syncop_lookup

To summarize-
I/O thread which acquired glfs_lock are waiting for epoll threads to receive response where as epoll threads are waiting for I/O threads to release lock. 

Similar issue was identified earlier (bug1693575).

There could be other issues at different layers depending on how client xlators choose to process these callbacks.

The correct way of avoiding or fixing these issues is to re-design upcall model which is to use different sockets for callback communication  instead of using same epoll threads. Raised github issue for that - https://github.com/gluster/glusterfs/issues/697 

Since it may take a while, raising this BZ to provide a workaround fix in gfapi layer for now 

Version-Release number of selected component (if applicable):


How reproducible:


Steps to Reproduce:
1.
2.
3.

Actual results:


Expected results:


Additional info:

--- Additional comment from Worker Ant on 2019-07-25 10:09:58 UTC ---

REVIEW: https://review.gluster.org/23107 (gfapi: Fix deadlock while processing upcall) posted (#1) for review on release-6 by soumya k

--- Additional comment from Worker Ant on 2019-07-25 10:16:57 UTC ---

REVIEW: https://review.gluster.org/23108 (gfapi: Fix deadlock while processing upcall) posted (#1) for review on master by soumya k

Comment 1 Worker Ant 2019-08-02 10:07:13 UTC
REVIEW: https://review.gluster.org/23151 (gfapi: Fix deadlock while processing upcall) posted (#1) for review on release-5 by soumya k

Comment 2 Worker Ant 2019-08-05 05:06:51 UTC
REVIEW: https://review.gluster.org/23151 (gfapi: Fix deadlock while processing upcall) merged (#1) on release-5 by soumya k


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