1
  2
  3
  4
  5
  6
  7
  8
  9
 10
 11
 12
 13
 14
 15
 16
 17
 18
 19
 20
 21
 22
 23
 24
 25
 26
 27
 28
 29
 30
 31
 32
 33
 34
 35
 36
 37
 38
 39
 40
 41
 42
 43
 44
 45
 46
 47
 48
 49
 50
 51
 52
 53
 54
 55
 56
 57
 58
 59
 60
 61
 62
 63
 64
 65
 66
 67
 68
 69
 70
 71
 72
 73
 74
 75
 76
 77
 78
 79
 80
 81
 82
 83
 84
 85
 86
 87
 88
 89
 90
 91
 92
 93
 94
 95
 96
 97
 98
 99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
use std::{
    ops::Deref,
    sync::{atomic::AtomicU64, Arc, RwLock},
};

use pagecache::FastMap8;

use super::*;

/// The `sled` embedded database!
#[derive(Clone)]
pub struct Db {
    context: Context,
    default: Arc<Tree>,
    tenants: Arc<RwLock<FastMap8<Vec<u8>, Arc<Tree>>>>,
}

unsafe impl Send for Db {}

unsafe impl Sync for Db {}

impl Deref for Db {
    type Target = Tree;

    fn deref(&self) -> &Tree {
        &self.default
    }
}

impl Db {
    /// Load existing or create a new `Db` with a default configuration.
    pub fn start_default<P: AsRef<std::path::Path>>(path: P) -> Result<Db> {
        let config = ConfigBuilder::new().path(path).build();
        Self::start(config)
    }

    /// Load existing or create a new `Db`.
    pub fn start(config: Config) -> Result<Db> {
        let _measure = Measure::new(&M.tree_start);

        let context = Context::start(config)?;

        let flusher_pagecache = context.pagecache.clone();
        let flusher = context.flush_every_ms.map(move |fem| {
            flusher::Flusher::new(
                "log flusher".to_owned(),
                flusher_pagecache,
                fem,
            )
        });
        *context._flusher.lock().unwrap() = flusher;

        // create or open the default tree
        let tx = context.pagecache.begin()?;
        let default = Arc::new(meta::open_tree(
            context.clone(),
            DEFAULT_TREE_ID.to_vec(),
            &tx,
        )?);

        let ret = Db {
            context: context.clone(),
            default,
            tenants: Arc::new(RwLock::new(FastMap8::default())),
        };

        let mut tenants = ret.tenants.write().unwrap();

        for (id, root) in context.pagecache.meta(&tx)?.tenants().into_iter() {
            let tree = Tree {
                tree_id: id.clone(),
                subscriptions: Arc::new(Subscriptions::default()),
                context: context.clone(),
                root: Arc::new(AtomicU64::new(root)),
            };
            tenants.insert(id, Arc::new(tree));
        }

        drop(tenants);

        #[cfg(feature = "event_log")]
        context.event_log.meta_after_restart(
            context
                .pagecache
                .meta(&tx)
                .expect("should be able to get meta under test")
                .clone(),
        );

        Ok(ret)
    }

    /// Open or create a new disk-backed Tree with its own keyspace,
    /// accessible from the `Db` via the provided identifier.
    pub fn open_tree<V: AsRef<[u8]>>(&self, name: V) -> Result<Arc<Tree>> {
        let name = name.as_ref();
        let tenants = self.tenants.read().unwrap();
        if let Some(tree) = tenants.get(name) {
            return Ok(tree.clone());
        }
        drop(tenants);

        let tx = self.context.pagecache.begin()?;

        let mut tenants = self.tenants.write().unwrap();
        let tree = Arc::new(meta::open_tree(
            self.context.clone(),
            name.to_vec(),
            &tx,
        )?);
        tenants.insert(name.to_vec(), tree.clone());
        drop(tenants);
        Ok(tree)
    }

    /// Remove a disk-backed collection.
    pub fn drop_tree(&self, name: &[u8]) -> Result<bool> {
        if name == DEFAULT_TREE_ID {
            return Err(Error::Unsupported(
                "cannot remove the core structures".into(),
            ));
        }
        trace!("dropping tree {:?}", name,);

        let mut tenants = self.tenants.write().unwrap();

        let tree = if let Some(tree) = tenants.remove(&*name) {
            tree
        } else {
            return Ok(false);
        };

        let tx = self.context.pagecache.begin()?;

        let mut root_id =
            Some(self.context.pagecache.meta_pid_for_name(&name, &tx)?);

        let leftmost_chain: Vec<PageId> = tree
            .path_for_key(b"", &tx)?
            .into_iter()
            .map(|(id, _frag, _tp)| id)
            .collect();

        loop {
            let res = self.context.pagecache.cas_root_in_meta(
                name.to_vec(),
                root_id,
                None,
                &tx,
            )?;

            if let Err(actual_root) = res {
                root_id = actual_root;
            } else {
                break;
            }
        }

        tree.root
            .store(u64::max_value(), std::sync::atomic::Ordering::SeqCst);

        // drop writer lock
        drop(tenants);

        tree.gc_pages(leftmost_chain)?;

        tx.flush();

        Ok(true)
    }

    /// Returns the trees names saved in this Db.
    pub fn tree_names(&self) -> Vec<Vec<u8>> {
        let tenants = self.tenants.read().unwrap();
        tenants.iter().map(|(name, _)| name.clone()).collect()
    }

    /// Returns `true` if the database was
    /// recovered from a previous process.
    /// Note that database state is only
    /// guaranteed to be present up to the
    /// last call to `flush`! Otherwise state
    /// is synced to disk periodically if the
    /// `sync_every_ms` configuration option
    /// is set to `Some(number_of_ms_between_syncs)`
    /// or if the IO buffer gets filled to
    /// capacity before being rotated.
    pub fn was_recovered(&self) -> bool {
        self.context.was_recovered()
    }

    /// Generate a monotonic ID. Not guaranteed to be
    /// contiguous. Written to disk every `idgen_persist_interval`
    /// operations, followed by a blocking flush. During recovery, we
    /// take the last recovered generated ID and add 2x
    /// the `idgen_persist_interval` to it. While persisting, if the
    /// previous persisted counter wasn't synced to disk yet, we will do
    /// a blocking flush to fsync the latest counter, ensuring
    /// that we will never give out the same counter twice.
    pub fn generate_id(&self) -> Result<u64> {
        self.context.generate_id()
    }
}