GracefulStop() may be blocked until all server-side streams
are finished. There is no control over such streams yet, so
application may be frozen in shutdown stage.
Naive solution is to add timeout for GracefulStop(). At this
point healthy connection will be finished and unhealthy
connections will be terminated by Stop().
Signed-off-by: Alex Vanin <alexey@nspcc.ru>
(cherry picked from commit 683439970a)
We already use thread-safe LRU and mutex shouldn't be taken while making
network requests.
Signed-off-by: Evgenii Stratonikov <evgeniy@nspcc.ru>
(cherry picked from commit 22b1208a20)
Similarly to `Get`. Also fix a bug where `ErrNotFound` is returned
instead of `ErrRangeOutOfBounds`.
Signed-off-by: Evgenii Stratonikov <evgeniy@nspcc.ru>
(cherry picked from commit 1fe9cd4d36)
Signed-off-by: Alex Vanin <alexey@nspcc.ru>
Metabase is expected to contain actual information about objects stored
in shard. If the object is present in metabase but is missing from
blobstor, peform an additional attempt to fetch it directly without
consulting metabase. Such a situation is unexpected, so error counter
is increased for the shard which has the object in the metabase. We
don't increase error counter for the shard which has the object in
blobstor, because some garbage can be expected there. In this
implementation there is no overhead for objects which are really
missing, i.e. are not present in any metabase.
Signed-off-by: Evgenii Stratonikov <evgeniy@nspcc.ru>
(cherry picked from commit 69e1e6ca20)
Signed-off-by: Alex Vanin <alexey@nspcc.ru>
If pre-existing blobovnicza is initialized, it's size should be updated
even if all buckets are in place.
Signed-off-by: Evgenii Stratonikov <evgeniy@nspcc.ru>
It was added back in 2fb379b7 when we had many shard modes. Now we have
only two and comments for constants are rather descriptive.
Signed-off-by: Evgenii Stratonikov <evgeniy@nspcc.ru>
It was broken since `initConfig` was made before every execution of the
command and flags have not been read by cobra yet, so it was impossible to
print config file path if `verbose` flag was set in command line not in
config file.
Signed-off-by: Pavel Karpy <carpawell@nspcc.ru>
In previous implementation IR incorrectly verified `SetEACL` event of
`Container` contract. The incorrect behavior could be reproduced in two
ways:
1. Create container using session, and perform `SetEACL` operation
with a key that is different from the session one.
2. Create container using session, and perform `SetEACL` w/o a
session, but sign it using session key from the `Put` operation.
The problem was in the `checkSetEACL` validation method of IR container
processor. It always used session token used for container creation
during session ownership check.
Signed-off-by: Leonard Lyubich <leonard@nspcc.ru>
We could also ignore errors during evacuate, but this requires
unmarshaling objects first which slowers the process considerably.
Signed-off-by: Evgenii Stratonikov <evgeniy@nspcc.ru>
