Whamcloud - gitweb
LU-16290 lnet: Signal completion on ping send failure 20/49020/3
authorChris Horn <chris.horn@hpe.com>
Tue, 1 Nov 2022 20:33:18 +0000 (14:33 -0600)
committerOleg Drokin <green@whamcloud.com>
Mon, 14 Nov 2022 08:26:47 +0000 (08:26 +0000)
commit48c34c71de65e8a251a218bc9ecb7c5ed522d786
treed7da95854c287c88802e0079c7b47cc3a20b5e6a
parent3dc261c06434eceee3ba9ef86d1f376954b2d234
LU-16290 lnet: Signal completion on ping send failure

Call complete() on the ping_data::completion if we get
LNET_EVENT_SEND with non-zero status. Otherwise the thread which
issued the ping is stuck waiting for the full ping timeout.

A pd_unlinked member is added to struct ping_data to indicate whether
the associated MD has been unlinked. This is checked by lnet_ping() to
determine whether it needs to explicitly called LNetMDUnlink().

Lastly, in cases where we do not receive a reply, we now return the
value of pd.rc, if it is non-zero, rather than -EIO. This can provide
more information about the underlying ping failure.

HPE-bug-id: LUS-11317
Test-Parameters: trivial testlist=sanity-lnet
Signed-off-by: Chris Horn <chris.horn@hpe.com>
Change-Id: I1bc573cf7397e319993fa8aabb31c5f3b59768e7
Reviewed-on: https://review.whamcloud.com/c/fs/lustre-release/+/49020
Tested-by: jenkins <devops@whamcloud.com>
Tested-by: Maloo <maloo@whamcloud.com>
Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com>
Reviewed-by: Frank Sehr <fsehr@whamcloud.com>
Reviewed-by: Oleg Drokin <green@whamcloud.com>
lnet/lnet/api-ni.c
lustre/tests/sanity-lnet.sh