Commits on Sep 8, 2020
-
LU-13553 lnd: gracefully handle unexpected events
When a tx completes kiblnd_tx_complete() callback is invoked. We ensure: LASSERT (tx->tx_sending > 0); However this assert is being triggered in some rare scenarios. The reason tx_sending would be 0 at this point is because: 1. ib_post_send() failed but OFED stack is still sending a tx complete event. 2. We're getting two different events for the same tx Instead of asserting, ignore that tx_complete event and print the tx pointer and its status. Lustre-change: https://review.whamcloud.com/38669 Lustre-commit: 60f9f53 Signed-off-by: Amir Shehata <ashehata@whamcloud.com> Change-Id: I8cd192538c0c80abaef23a4b6e6906936043060b Reviewed-by: Andreas Dilger <adilger@whamcloud.com> Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com> Signed-off-by: Minh Diep <mdiep@whamcloud.com> Reviewed-on: https://review.whamcloud.com/38752 Tested-by: jenkins <devops@whamcloud.com> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Oleg Drokin <green@whamcloud.com> -
LU-13600 ptlrpc: limit rate of lock replays
Clients send all lock replays at once and that may overwhelm server with huge amount of replays in recovery queue causing OOM effects. Patch adds rate control for lock replays on client. Patch includes also later fix for signal_completed_replay() race. Lustre-change: https://review.whamcloud.com/38920 Lustre-commit: 3b613a4 Lustre-change: https://review.whamcloud.com/39140 Lustre-commit: dc654756af63bd30802ebd86074019d1533a4d8f Signed-off-by: Mikhail Pershin <mpershin@whamcloud.com> Change-Id: Ie557f8481c5facb690468d7136cf5feebe4e8f11 Reviewed-on: https://review.whamcloud.com/39111 Tested-by: jenkins <devops@whamcloud.com> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
-
LU-13088 ldlm: Fix sleeping function called in atomic
target_recovery_overseer() can sleep while holding a spinlock, which triggers a BUG warning. It is easily fixed by dropping the spinlock before waiting. In the case where the task waits, no useful information that could be protected by the spinlock is held, so nothing can be lost by dropping it. Lustre-change: https://review.whamcloud.com/#/c/37063/ Lustre-commit: b29b931 Signed-off-by: Mr NeilBrown <neilb@suse.de> Change-Id: I8bb3d02523b5dcfadac19f01ccb736d7b7f28239 Reviewed-on: https://review.whamcloud.com/37063 Tested-by: jenkins <devops@whamcloud.com> Reviewed-by: James Simmons <jsimmons@infradead.org> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Mike Pershin <mpershin@whamcloud.com> Reviewed-by: Oleg Drokin <green@whamcloud.com> Reviewed-on: https://review.whamcloud.com/39283
-
LU-12424 lnet: prevent loop in LNetPrimaryNID()
If discovery is disabled locally or at the remote end, then attempt discovery only once. Do not update the internal database when discovery is disabled and do not repeat discovery. This change prevents LNet from getting hung waiting for discovery to complete. Signed-off-by: Amir Shehata <ashehata@whamcloud.com> Change-Id: I4543b0f71e6cf297a1a5f058ebcc6bf74b8ac328 Reviewed-on: https://review.whamcloud.com/35191 Reviewed-by: Olaf Weber <olaf.weber@hpe.com> Tested-by: Jenkins Reviewed-by: Chris Horn <hornc@cray.com> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Oleg Drokin <green@whamcloud.com> Reviewed-on: https://review.whamcloud.com/38890 Reviewed-by: Chris Horn <chris.horn@hpe.com> Tested-by: jenkins <devops@whamcloud.com>
-
LU-12222 lnet: Introduce constant for the lolnd NID
This patch adds a new constant, LNET_NID_LO_0, to represent the lolnd NID 0@lo. Lustre-change: https://review.whamcloud.com/38312 Lustre-commit: 56203e4 HPE-bug-id: LUS-8457 Signed-off-by: Chris Horn <hornc@cray.com> Change-Id: I3e57637f297b8de306905a447af8f025e31d1fcf Reviewed-on: https://review.whamcloud.com/38863 Tested-by: jenkins <devops@whamcloud.com> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com> Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
-
LU-12222 lnet: Primary NID of lolnd NID is the lolnd NID
We want Lustre traffic that is intended for the local peer to be sent and received over the lolnd. The function ptlrpc_uuid_to_peer() will currently resolve a NID to the lolnd NID, but ptlrpc_connection_get() will overwrite this selection with the result from LNetPrimaryNID(). Have LNetPrimaryNID return the lolnd NID when it is passed the lolnd NID. Lustre-change: https://review.whamcloud.com/38313 Lustre-commit: 33d2e44 HPE-bug-id: LUS-8457 Signed-off-by: Chris Horn <hornc@cray.com> Change-Id: I02708bb45f8440091782ca7886bac7656efb0223 Reviewed-on: https://review.whamcloud.com/38864 Tested-by: jenkins <devops@whamcloud.com> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com> Reviewed-by: Andreas Dilger <adilger@whamcloud.com>
-
LU-12222 ptlrpc: Check if NID is local, not just lolnd NID
There's a couple places where we check whether a NID is the lolnd NID but we really want to know whether the NID is local. Use LNetIsPeerLocal() to accomplish this. Lustre-change: https://review.whamcloud.com/38388 Lustre-commit: 95bcc24 Signed-off-by: Chris Horn <hornc@cray.com> Change-Id: Ia17b9b4b54fd1063c42a6f8bdd0e593be1086683 Reviewed-on: https://review.whamcloud.com/38865 Tested-by: jenkins <devops@whamcloud.com> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com> Reviewed-by: Andreas Dilger <adilger@whamcloud.com> Reviewed-by: Oleg Drokin <green@whamcloud.com>
-
LU-9971 lnet: use after free in lnet_discover_peer_locked()
When the lnet_net_lock is unlocked, the peer attached to an lnet_peer_ni (found via lnet_peer_ni::lpni_peer_net->lpn_peer) can change, and the old peer deallocated. If we are really unlucky, then all the churn could give us a new, different, peer at the same address in memory. Change the reference counting on the lnet_peer lp so that it is guaranteed to be alive when we relock the lnet_net_lock for the cpt. When the reference count is dropped lp may go away if it was unlinked, but the new peer is guaranteed to have a different address, so we can still correctly determine whether the peer changed and discovery should be redone. LU-9971 lnet: fix peer ref counting Exit from the loop after peer ref count has been incremented to avoid wrong ref count. The code makes sure that a peer is queued for discovery at most once if discovery is disabled. This is done to use discovery as a standard ping for gateways which do not have discovery feature or discovery is disabled. Signed-off-by: Olaf Weber <olaf.weber@hpe.com> Change-Id: Ia44dce20074b27ec0e77d7c1908c6a44ec73d326 Reviewed-on: https://review.whamcloud.com/28944 Reviewed-by: Amir Shehata <ashehata@whamcloud.com> Tested-by: Jenkins Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: James Simmons <uja.ornl@yahoo.com> Reviewed-by: Oleg Drokin <green@whamcloud.com> Reviewed-on: https://review.whamcloud.com/38891 Tested-by: jenkins <devops@whamcloud.com> Reviewed-by: Chris Horn <chris.horn@hpe.com> Reviewed-by: James Simmons <jsimmons@infradead.org>
-
LU-13278 lnet: Reconcile discovery push and reply handling
Reconcile the logic for updating the multi-rail flag of a peer when processing a discovery PUSH with the logic used when processing a discovery REPLY. Cray-bug-id: LUS-8516 Signed-off-by: Chris Horn <hornc@cray.com> Change-Id: Idfb4c3729822d03b71f9440ac66176ae6b886022 Reviewed-on: https://review.whamcloud.com/37674 Tested-by: jenkins <devops@whamcloud.com> Tested-by: Maloo <maloo@whamcloud.com> Reviewed-by: Shaun Tancheff <shaun.tancheff@hpe.com> Reviewed-by: Serguei Smirnov <ssmirnov@whamcloud.com> Reviewed-by: Stephen Champion <stephen.champion@hpe.com> Reviewed-by: Oleg Drokin <green@whamcloud.com> Reviewed-on: https://review.whamcloud.com/39575 Reviewed-by: Chris Horn <chris.horn@hpe.com>
-
LU-13763 osc: don't allow negative grants
Add check in the osc_init_grant() to prevent possible underflow of cl_avail_grant and report error if it happens Cherry-picked-from: c84fdeb5469bdc507caf1b4d5b876fce47cf5d86 Cherry-picked-from-change: https://review.whamcloud.com/#/c/39380 Cherry-picked-from-patch: 2 Cherry-picked-from-branch: b2_12 Cherry-picked-from-status: Maloo +1, Review +1 Signed-off-by: Mikhail Pershin <mpershin@whamcloud.com> Change-Id: Idcd25ed427c23735e1cdc70359bace43b5b9d886
-
LU-12687 osc: consume grants for direct I/O
New IO engine implementation lost consuming grants by direct I/O writes. That led to early emergence of out of space condition during direct I/O. The below illustrates the problem: # OSTSIZE=100000 sh llmount.sh # dd if=/dev/zero of=/mnt/lustre/file bs=4k count=100 oflag=direct dd: error writing ‘/mnt/lustre/file’: No space left on device Consume grants for direct I/O. Try to consume grants in osc_queue_sync_pages() when it is called for pages which are being writted in direct i/o. Tests are added to verify grant consumption in buffered and direct i/o and to verify direct i/o overwrite when ost is full. The overwrite test is for ldiskfs only as zfs is unable to overwrite when it is full. Cherry-picked-from: 8dd02c50f7df6fd0af9ead75d5d7774f32c211e2 Cherry-picked-from-change: https://review.whamcloud.com/#/c/39386 Cherry-picked-from-patch: 10 Cherry-picked-from-branch: b2_12 Cherry-picked-from-status: Maloo +1, Review +1 Lustre-change: https://review.whamcloud.com/35896 Lustre-commit: 05f326a Fixes: 9fe4b52 ("LU-1030 osc: new IO engine implementation") Signed-off-by: Vladimir Saveliev <c17830@cray.com> Change-Id: I9a199452c564e8e8ad02f79231e8481166f3666e Cray-bug-id: LUS-7036
-
LU-13089 osc: revert "glimpse - search for active lock"
Revert "LU-11670 osc: glimpse - search for active lock" This could cause assertion failures like below: LustreError: 13759:0:(ldlm_lock.c:213:ldlm_lock_put()) ASSERTION((((( lock))->l_flags & (1ULL << 50)) != 0) ) failed: LustreError: 10188:0:(ldlm_lock.c:205:ldlm_lock_put()) ASSERTION( atomic_read(&lock->l_refc) > 0 ) failed: LustreError: 10188:0:(ldlm_lock.c:205:ldlm_lock_put()) LBUG A glimpse cb race with cancel cb. This reverts commit 2548cb9 Conflicts: lustre/tests/sanityn.sh Cherry-picked-from: 37205c60bb2d99363a0c9dbf29d8f4fd684b6fab Cherry-picked-from-change: https://review.whamcloud.com/#/c/39819 Cherry-picked-from-patch: 1 Cherry-picked-from-branch: b2_12 Cherry-picked-from-status: Maloo +1, No Review Cherry-picked-from-notes: Removed test_103 Signed-off-by: Bobi Jam <bobijam@whamcloud.com> Change-Id: I12063d0b3f1411e0d44393823a3e220cea6567d5
