Do not sync peer OSDs before listing
Sync before listing was added to wait for all PG writes possibly left in queue from the previous master to finish before listing it But in fact it may block the cluster when EC is used and some unstable writes are left in the queue - they block journal flushing, rollback/stabilize is required to unblock them, but rollback/stabilize may only happen after PG is peered. But peering needs listings, listings are requested only after sync, and sync itself waits for currently blocked writes waiting in the queue
This commit is contained in:
@@ -583,7 +583,7 @@ void blockstore_impl_t::process_list(blockstore_op_t *op)
|
||||
replace_stable(dirty_it->first.oid, 0, clean_stable_count, stable_count, stable);
|
||||
}
|
||||
}
|
||||
else if (IS_STABLE(dirty_it->second.state))
|
||||
else if (IS_STABLE(dirty_it->second.state) || (dirty_it->second.state & BS_ST_INSTANT))
|
||||
{
|
||||
// First try to replace a clean stable version in the first part of the list
|
||||
if (!replace_stable(dirty_it->first.oid, dirty_it->first.version, 0, clean_stable_count, stable))
|
||||
|
||||
Reference in New Issue
Block a user