# drbd-9.3.4-rc.1 **URL:** \ **Category:** Release Announcements\ **Tags:** drbd\ **Created:** [September 8, 2026, 7:45pm UTC](https://forums.linbit.com/t/drbd-9-3-4-rc-1/1267 "2026-09-08T19:45:18Z")\ **Posts on this page:** 1\ **Page:** 1
**Author:** ![phil\_reisner](https://avatars.discourse-cdn.com/v4/letter/p/779978/32.png) [@phil\_reisner](https://forums.linbit.com/u/phil_reisner)\ **Post date:** [September 8, 2026, 7:45pm UTC](https://forums.linbit.com/t/drbd-9-3-4-rc-1/1267/1 "2026-09-08T19:45:18Z")
Hi, Following yesterday's 9.2.20-rc.1 release, here is the release in the 9.3 series. The delay was, as so often, due to issues in the kernel compatibility system. Supporting such a broad range of kernel versions is challenging. Solved for this time. This is a release candidate. Please test it. 9.3.4-rc.1 (api:genl2/proto:86-101,118-124/transport:22) -------- \* Generate the netlink interface code with the upstream YNL generator    instead of the genl\_magic macro system \* Fix a ProtocolError disconnect loop ("BAD! BarrierAck received with    n\_writes=...") when two volumes of one connection replicate and    resync without replication at the same time \* Tolerate a verify-alg mismatch when connecting, and refuse to start    an online verify while the algorithms differ, instead of rejecting    the connection outright \* Fix several bitmap lifetime bugs around attach and detach: a NULL    bitmap dereference in the peer device statistics, a missing bitmap    in drbd\_set\_sync(), concurrent drbd\_bm\_free() callers, and freeing    a bitmap that a later attach published \* Fix a device reference leak and a NULL pointer dereference when a    peer request's bio allocation fails under memory pressure \* Hold state\_rwlock while walking the transfer log from debugfs \* All fixes from 9.2.20   - Fix a sync target overwriting acknowledged writes with older resync     data when the sync source lost the diskless primary; such writes are     now held back or postponed (new feature flag WRITE\_POSTPONE) instead     of being silently rolled back   - Fix data corruption when a peer bitmap that is being cleared by a     running resync was copied for a second peer   - Fix several ways a Primary lost its obligation to start a new data     generation after losing a peer, each ending in silent divergence,     and a spurious I/O error when a deferred generation resumes; the     obligation is now tracked per volume and logged   - Fix a diskless Primary losing a peer: rotate the current UUID so the     stale peer resyncs on return, outdate far-away members before     resuming I/O, and keep the sole UpToDate survivor UpToDate; also     fixes promotions freezing I/O on nodes not connected to every member   - Fix a resync target ending with stale data or an old data generation     when a crashed primary rejoins or a diskless primary announces a new     generation mid-resync   - Fix al-updates=no: honour it at attach time, and mark all bitmap     slots out of sync after a primary crash   - Fix kernel memory corruption by a malicious or man-in-the-middle     peer sending out-of-range payload, drain or digest sizes, and a     divide error caused by bogus peer block sizes   - Fix a use-after-free of the resource when a connections dump is     interrupted, and a refcount imbalance flooding the log with put\_ldev     assertions after a resync request allocation failure   - Fix soft lockups: the sender thread pinning a CPU while its     connection is down, and IRQ-disabled bitmap scans on multi-TiB     volumes   - Fix several deadlocks and hangs: whole-bitmap IO vs clear\_peer\_slot()     or online resize, WFBitMapS on an equal-UUID reconnect with leftover     out-of-sync bits, demotion starved by a failing connect handshake,     and connection teardown blocked behind an unacknowledged resync     conflict   - Fix many two-phase-commit issues: stale replies answering the wrong     transaction, prepared peers that disconnected, forwarding node     handling, a state\_sem leak on timeout, and unbounded connect     retries; log which volume a connect is waiting for   - Fix resyncs stalling short of completion, and a connection dropping     to StandAlone when a crashed primary rejoins during a resync   - Fix drbd\_suspend\_io() not waiting for in-flight writes; a detach     hanging on a dead backing device stays interruptible   - Fix restoring quorum after a reboot when the remaining members are     diskless   - Fix connection losses: online resize ending in ProtocolError,     volume-less connections reconnecting every ping-timeout, lb-tcp     "drbdadm down" stalling for a connect-int, and a TLS control socket     close going unnoticed   - Declare a connection broken where a two-phase-commit send failed,     and repeat a disconnect without delay and with a bound; together     they end a hanging "drbdadm disconnect"   - Do not account requests as waiting for the activity log while it is     suspended; the stale counters throttled resync to c-min-rate and     drained the epoch for every incoming peer write [https://pkg.linbit.com//downloads/drbd/9/drbd-9.3.4-rc.1.tar.gz](https://pkg.linbit.com//downloads/drbd/9/drbd-9.3.4-rc.1.tar.gz) > -Phil