[pbs-devel] [PATCH proxmox-backup v2] datastore: simplify update-datastore-cache socket command
Hannes Laimer
h.laimer at proxmox.com
Thu Aug 22 15:58:26 CEST 2024
Thanks for the feedback I'll send a v3! One comment inline
On Thu Aug 22, 2024 at 9:32 AM CEST, Wolfgang Bumiller wrote:
> On Mon, Aug 19, 2024 at 11:45:22AM GMT, Hannes Laimer wrote:
> > ... by always dropping the cache entry and not allowing it to be
> > re-added based on the maintenance mode. The check is only needed for
>
> What "check are you referring to here?
>
> > `Lookup` operations since lookups with those wouldn't fail based on
> > maintenance mode alone. Already running tasks won't be affected and
> > since they're the only ones holding a reference to a DataStore, it,
> > and the file handles it holds will be dropped automatically once the
> > (last) task finishes.
> >
> > Signed-off-by: Hannes Laimer <h.laimer at proxmox.com>
> > ---
> > v2:
> > - rebase onto master
> >
> > pbs-datastore/src/datastore.rs | 26 ++++++++++----------------
> > src/api2/config/datastore.rs | 7 +++----
> > src/bin/proxmox-backup-proxy.rs | 6 +++---
> > 3 files changed, 16 insertions(+), 23 deletions(-)
> >
> > diff --git a/pbs-datastore/src/datastore.rs b/pbs-datastore/src/datastore.rs
> > index d0f3c53a..69b4c83e 100644
> > --- a/pbs-datastore/src/datastore.rs
> > +++ b/pbs-datastore/src/datastore.rs
> > @@ -183,10 +183,15 @@ impl DataStore {
> > )?)
> > };
> >
> > + let is_offline = config
> > + .get_maintenance_mode()
> > + .map_or(false, |m| m.is_offline());
> > let datastore = DataStore::with_store_and_config(chunk_store, config, Some(digest))?;
> >
> > let datastore = Arc::new(datastore);
> > - datastore_cache.insert(name.to_string(), datastore.clone());
> > + if !is_offline {
> > + datastore_cache.insert(name.to_string(), datastore.clone());
> > + }
>
> So previously we always only ever had 1 DataStore+ChunkStore pair. Now,
> we can have multiple instances for an offline data store, as well as
> after receiving an "update-datastore-cache" command, which now forces
> multiple instances.
>
> This seems to go against the original idea here. Why do we want this?
>
This is the check referenced in the commit message.
The idea was to avoid having not-counted references that hold file
handles. But just not adding it to the cache does not stop it from
holding the file handle, so this indeed does not make sense in its
current form.
As we've discussed off-list I'll just give the DataStore ref returned
for Lookup operations a "dummy"-ChunkStore for now
> >
> > if let Some(operation) = operation {
> > update_active_operations(name, operation, 1)?;
> > @@ -208,21 +213,10 @@ impl DataStore {
> > Ok(())
> > }
> >
> > - /// trigger clearing cache entry based on maintenance mode. Entry will only
> > - /// be cleared iff there is no other task running, if there is, the end of the
> > - /// last running task will trigger the clearing of the cache entry.
> > - pub fn update_datastore_cache(name: &str) -> Result<(), Error> {
> > - let (config, _digest) = pbs_config::datastore::config()?;
> > - let datastore: DataStoreConfig = config.lookup("datastore", name)?;
> > - if datastore
> > - .get_maintenance_mode()
> > - .map_or(false, |m| m.is_offline())
> > - {
> > - // the datastore drop handler does the checking if tasks are running and clears the
> > - // cache entry, so we just have to trigger it here
> > - let _ = DataStore::lookup_datastore(name, Some(Operation::Lookup));
> > - }
> > -
> > + /// removes datastore from cache
> > + pub fn drop_from_datastore_cache(name: &str) -> Result<(), Error> {
>
> There's no remaining error case here, so we wouldn't need a `Result`.
>
> > + let mut datastore_cache = DATASTORE_MAP.lock().unwrap();
> > + datastore_cache.remove(name);
> > Ok(())
> > }
> >
> > diff --git a/src/api2/config/datastore.rs b/src/api2/config/datastore.rs
> > index ca6edf05..65d88265 100644
> > --- a/src/api2/config/datastore.rs
> > +++ b/src/api2/config/datastore.rs
> > @@ -396,16 +396,15 @@ pub fn update_datastore(
> > data.tuning = update.tuning;
> > }
> >
> > - let mut maintenance_mode_changed = false;
> > + let mut drop_store_from_cache = false;
> > if update.maintenance_mode.is_some() {
> > - maintenance_mode_changed = data.maintenance_mode != update.maintenance_mode;
> > -
> > let maintenance_mode = match update.maintenance_mode {
> > Some(mode_str) => Some(MaintenanceMode::deserialize(
> > proxmox_schema::de::SchemaDeserializer::new(mode_str, &MaintenanceMode::API_SCHEMA),
> > )?),
> > None => None,
> > };
> > + drop_store_from_cache = maintenance_mode.as_ref().map_or(false, |m| m.is_offline());
>
> (This could be set in the `Some` match arm more efficiently as it
> already defaults to false)
>
> > data.set_maintenance_mode(maintenance_mode)?;
> > }
> >
> > @@ -420,7 +419,7 @@ pub fn update_datastore(
> > }
> >
> > // tell the proxy it might have to clear a cache entry
>
> The comment is now wrong.
>
> > - if maintenance_mode_changed {
> > + if drop_store_from_cache {
> > tokio::spawn(async move {
> > if let Ok(proxy_pid) =
> > proxmox_rest_server::read_pid(pbs_buildcfg::PROXMOX_BACKUP_PROXY_PID_FN)
> > diff --git a/src/bin/proxmox-backup-proxy.rs b/src/bin/proxmox-backup-proxy.rs
> > index 041f3aff..04622d43 100644
> > --- a/src/bin/proxmox-backup-proxy.rs
> > +++ b/src/bin/proxmox-backup-proxy.rs
> > @@ -274,11 +274,11 @@ async fn run() -> Result<(), Error> {
> > Ok(Value::Null)
> > })?;
> >
> > - // clear cache entry for datastore that is in a specific maintenance mode
> > + // clear cache entry for datastore
> > command_sock.register_command("update-datastore-cache".to_string(), |value| {
> > if let Some(name) = value.and_then(Value::as_str) {
> > - if let Err(err) = DataStore::update_datastore_cache(name) {
> > - log::error!("could not trigger update datastore cache: {err}");
> > + if let Err(err) = DataStore::drop_from_datastore_cache(name) {
> > + log::error!("could not drop datastore from cache: {err}");
> > }
> > }
> > Ok(Value::Null)
> > --
> > 2.39.2
More information about the pbs-devel
mailing list