we try to load the correct media in a loop until we find the correct tape.
when encountering an error or wrong tape, we want to log that (and send
an email if one is set) that requests the correct tape.
while trying to avoid printing the same errors more than once in a row,
we had at least one case (starting with an empty tape in the drive)
which would not print/send any tape request.
reworking that code to use a custom 'TapeRequest' enum, which contains
the state + error message, and a helper that prints and sends an email
when the state changes
this reduces the change check/log to a single variable, instead of 4
(tried, last_media_uuid, last_error, failure_reason)
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
Add test code to the first locate_file command, compute locate_offset.
Subsequent locate_file commands use that offset.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
we have a static list of filesystems and their capabilities regarding
file attributes and fs features (e.g. sockets/fifos/etc) which also
includes xattrs,acls and fcaps
if we did not know a filesystem by its magic number (for example cephfs),
we did not even attempt to read xattrs, etc.
this patch adds those flags by default to unknown filesystems, and
removes them when we encounter EOPNOTSUPP (to remove the number
of syscalls)
with this, we should be able to catch xattrs/acls/fcaps on all
(unknown) fs types that support them
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
These require mounting using the regular 'mount' syscall.
Auto-generates an appropriate mount path.
Note that subvols with mountpoint=none cannot be mounted this way, and
would require setting the mountpoint property, which is not possible as
the zpools have to be imported with readonly=on.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Uses the ZFS utils to detect, import and mount zpools. These are
available as a new Bucket type 'zpool'.
Requires some minor changes to the existing disk and partiton detection
code, so the ZFS-specific part can use the information gathered in the
previous pass to associate drive names with their 'drive-xxxN.img.fidx'
node.
For detecting size, the zpool has to be imported. This is only done with
pools containing 5 or less disks, as anything else might take too long
(and should be seldomly found within VMs).
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Even through best efforts at keeping it small, including the ZFS tools
in the initramfs seems to have exhausted the small overhead we had left
- give it a bit more RAM to compensate.
Also disable the ZFS ARC, as it's no use in such a memory constrained
environment, and we cache on the QEMU/rust layer anyway.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
The future needs to be removed from the pending map in any case, even if
it returned an error, else all upcoming calls to access this key will
always return the same error.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Signed-off-by: Wolfgang Bumiller <w.bumiller@proxmox.com>
sort the chunks we want to backup to tape by inode, to gain some
speed on spinning disks. this is done per index, not globally.
costs a bit memory, but not too much, about 16 bytes per chunk which
would mean ~4MiB for a 1TiB index with 4MiB chunks.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
so that we can reuse that information
the removal of the adding to the corrupted list is ok, since
'get_chunks_in_order' returns them at the end of the list
and we do the same if the loading fails later in 'verify_index_chunks'
so we still mark them corrupt
(assuming that the load will fail if the stat does)
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
since the output:
Result: "<UPID>"
is not really interesting, show instead the task log while
the datastore is creating, since it is now run in a worker
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Setting this to 0 is not just useless, but breaks the logic horribly
enough to cause random segfaults - better forbid this, to avoid someone
else having to debug it again ;)
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
admin/datastore reads linearly only, so no need for cache (capacity of 1
basically means no cache except for the currently active chunk).
mount can do random access too, so cache last 8 chunks for possibly a
mild performance improvement.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Implemented as a seperate struct SeekableCachedChunkReader that contains
the original as an Arc, since the read_at future captures the
CachedChunkReader, which would otherwise not work with the lifetimes
required by AsyncRead. This is also the reason we cannot use a shared
read buffer and have to allocate a new one for every read. It also means
that the struct items required for AsyncRead/Seek do not need to be
included in a regular CachedChunkReader.
This is intended as a replacement for AsyncIndexReader, so we have less
code duplication and can utilize the LRU cache there too (even though
actual request concurrency is not supported in these traits).
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Supports concurrent 'access' calls to the same key via a
BroadcastFuture. These are stored in a seperate HashMap, the LruCache
underneath is only modified once a valid value has been retrieved.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
Explicitly test that data will stay available and can be retrieved
immediately via listen(), even if the future producing the data and
notifying the consumers was already run in the past.
Wasn't broken or anything, but helps with understanding IMO.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
in PVE, the logic how wearout gets read from the smartctl output was
changed from a vendor -> id map to a sorted list of specific
attribute field names.
copy that list to pbs (in the same order), and use that to get the
wearout
in the future we might want to split the disk logic into its own crate
and reuse it in pve
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
we skip snapshots that are older than the newest snapshot of the group in
the target datastore, log it so the user can know why it is not synced
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
so that longer running creates (e.g. a slow storage), does not
run in a timeout and we can follow its creation
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
when we remove a datastore via api/cli, the proxy
has sometimes leftover references to that datastore in its
DATASTORE_MAP which includes an open filehandle on the
'.lock' file
this prevents unmounting/exporting the datastore even after removal,
only a reload/restart of the proxy did help
add a command to our command socket, which removes all non
configured datastores from the map, dropping the open filehandle
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Dietmar Maurer <dietmar@proxmox.com>
by implementing a custom error type that is either 'TimeOut' or
'Other'.
In the api, check in the worker loop for exactly 'TimeOut' errors and continue only
then. All other errors lead to a aborted task.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
removing the backup dir must acquire the snapshot lock, else it can
happen that we remove a snapshot while it is being restored
or backed up to tape
the original commit that adds the force flag
(c9756b40d1)
mentions that the prune checks itself if the snapshot is in use,
but i could not find such code, so simply set force to false
to avoid failing and aborting the prune job, warn if it could not
and continue
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
This reverts commit 75f9f40922, which is
no longer needed now that we use tokio >= 1.6 which contains the proper
fix.
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
this is deprecated with rustc 1.52+, and will become a hard error at
some point:
https://github.com/rust-lang/rust/issues/79202
Signed-off-by: Fabian Grünbichler <f.gruenbichler@proxmox.com>
from the SspDataEncryptionCapabilityPage
it seems we do not need it, since the EXTDECC flag is only used for
determining if the drive is capable to be configured via
ADI (Automation/Drive Interface) which we do not use at all.
this makes the call work with LTO-4 again
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
we want a 'media-set' selector in the gui, this makes it
very easy to do and is not as costly as reusing the media list,
since we do not need to iterate over all media (e.g. unassigned)
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
by extracting them via the api macro into the function signature
this fixes an issue, where giving 'since' and 'until' where not
used since we tried to extract them as 'str' while they were numbers.
Signed-off-by: Dominik Csapak <d.csapak@proxmox.com>
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
While the issue with vsock packets starving kernel memory is mostly
worked around by the '64k -> 4k buffer' patch in
'proxmox-backup-restore-image', let's be safe and also limit the number
of concurrent transfers. 8 downloads per VM seems like a fair value.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
The extract API call may be active for more than the watchdog timeout,
so a simple ping is not enough.
This adds an "inhibit" API, which will stop the watchdog from completing
as long as at least one WatchdogInhibitor instance is alive. Keep one in
the download task, so it will be dropped once it completes (or errors).
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
See this PR for more info: https://github.com/tokio-rs/tokio/pull/3756
As a workaround use a pair of connected unix sockets - this obviously
incurs some overhead, albeit not measureable on my machine. Once tokio
includes the fix we can go back to a DuplexStream for performance and
simplicity.
Signed-off-by: Stefan Reiter <s.reiter@proxmox.com>
