The first rotation is normally the one still opened by one or more
processes for writing, so it must NOT be replaced, removed, ..., as
this then makes the remaining logging, until those processes are
noticed that they should reopen the logfile due to rotation, goes
into nirvana, which is far from ideal for a log.
Only rotating (renaming) is OK for this active file, as this does not
invalidates the file and keeps open FDs intact.
So start compressing with the second rotation, which should be clear
to use, as all writers must have been told to reopen the log during
the last rotation, reopen is a fast operation and normally triggered
at least day ago (at least if one did not dropped the state file
manually), so we are fine to archive that one for real.
If we plan to allow faster rotation the whole rotation+reopen should
be locked, so that we can guarantee that all writers switched over,
but this is unlikely to be needed.
Again, this is was logrotate sanely does by default since forever.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
this is not the job of logrotate, and the real 20+ years battle
tested log rotate binary does not do so either as it's actually
pretty dangerous.
If we "replace" the file we break any logger which already opened a
new one here, e.g., a dameon starting up, and thus that writer would
log to nirvana.
It's the job of a logger to create a file if not existing, it makes
no sense to do it here.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
and use that in ApiConfig to avoid that it is owned by root if the
proxmox-backup-api process creates it first.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
for now log auth errors also to the syslog, on a protected (LAN
and/or firewalled) setup this should normally happen due to
missconfiguration, not tries to break in.
This reduces syslog noise *a lot*. A current full journal output from
the current boot here has 72066 lines, of which 71444 (>99% !!) are
"successful auth for user ..." messages
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
reuse the FileLogger module in append mode.
As it implements write, which is not thread safe (mutable self) and
we use it in a async context we need to serialize access using a
mutex.
Try to use the same format we do in pveproxy, namely the one which is
also used in apache or nginx by default.
Use the response extensions to pass up the userid, if we extract it
from a ticket.
The privileged and unprivileged dameons log both to the same file, to
have a unified view, and avoiding the need to handle more log files.
We avoid extra intra-process locking by reusing the fact that a write
smaller than PIPE_BUF (4k on linux) is atomic for files opened with
the 'O_APPEND' flag. For now the logged request path is not yet
guaranteed to be smaller than that, this will be improved in a future
patch.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Add a generous limit now and return the correct error (414 URI Too
Long). Otherwise we could to pretty larger GET requests, 64 KiB and
possible bigger (at 64 KiB my simple curl test failed due to
shell/curl limitations).
For now allow a 3072 characters as combined length of URI path and
query.
This is conform with the HTTP/1.1 RFCs (e.g., RFC 7231, 6.5.12 and
RFC 2616, 3.2.1) which do not specify any limits, upper or lower, but
require that all server accessible resources mus be reachable without
getting 414, which is normally fulfilled as we have various length
limits for stuff which could be in an URI, in place, e.g.:
* user id: max. 64 chars
* datastore: max. 32 chars
The only known problematic API endpoint is the catalog one, used in
the GUI's pxar file browser:
GET /api2/json/admin/datastore/<id>/catalog?..&filepath=<path>
The <path> is the encoded archive path, and can be arbitrary long.
But, this is a flawed design, as even without this new limit one can
easily generate archives which cannot be browsed anymore, as hyper
only accepts requests with max. 64 KiB in the URI.
So rather, we should move that to a GET-as-POST call, which has no
such limitations (and would not need to base32 encode the path).
Note: This change was inspired by adding a request access log, which
profits from such limits as we can then rely on certain atomicity
guarantees when writing requests to the log.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
needs new proxmox dependency to get the RpcEnvironment changes,
adding client_ip getter and setter.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
The 'Ok::<_, Self::Error>(res)' type annotation was from a time where
we could not use async, and had a combinator here which needed
explicity type information. We switched over to async in commit
91e4587343 and, as the type annotation
is already included in the Future type, we can safely drop it.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
This is only acquired in those two methods, both as shared. So it has
no use.
It seems, that it was planned in the past that the index deletion
should take the exclusive, while read and write takes the shared
flock on the index, as one can guess from the lock comments in commit
0465218953
But then later, in commit c8ec450e37)
the documented semantics where changed to use a temp file and do an
atomic rename instead for atomicity.
The reader shared flock on the index file was done inbetween,
probably as preparatory step, but was not removed again when strategy
was changed to using the file rename instead.
Do so now, to avoid confusion of readers and a useless flock.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Previously only Datastore.Modify was required for creating a new
datastore.
But, that endpoint allows one to pass an arbitrary path, of which all
parent directories will be created, this can allow any user with the
"Datastore Admin" role on "/datastores" to do some damage to the
system. Further, it is effectively a side channel for revealing the
systems directory structure through educated guessing and error
handling.
Add a new privilege "Datastore.Allocate" which, for now, is used
specifically for the create datastore API endpoint.
Add it only to the "Admin" role.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
avoiding the need for reshuffling all bits when a new privilege is
added at the start or in the middle of this definition.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
while we probably do not add much more to them, it still looks ugly.
If this was made so that adding a World readable API call is "hard"
and not done by accident, it rather should be done as a test on build
time. But, IMO, the API permission schema definitions are easy to
review, and not often changed/added - so any wrong World readable API
call will normally still caught.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
This can slow things down by a lot on setups with (relatively) high
seek time, in the order of doubling the backup times if cache isn't
populated with the last backups chunk inode info.
Effectively there's nothing known this protects us from in the
codebase. The only thing which was theorized about was the case
where a really long running backup job (over 24 hours) is still
running and writing new chunks, not indexed yet anywhere, then an
update (or manual action) triggers a reload of the proxy. There was
some theory that then a GC in the new daemon would not know about the
oldest writer in the old one, and thus use a less strict atime limit
for chunk sweeping - opening up a window for deleting chunks from the
long running backup.
But, this simply cannot happen as we have a per datastore process
wide flock, which is acquired shared by backup jobs and exclusive by
GC. In the same process GC and backup can both get it, as it has a
process locking granularity. If there's an old daemon with a writer,
that also has the lock open shared, and so no GC in the new process
can get exclusive access to it.
So, with that confirmed we have no need for a "half-assed"
verification in the backup finish step. Rather, we plan to add an
opt-in "full verify each backup on finish" option (see #2988)
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
fixed-width navi/toc links were not switched in color for small width
displays, and thus they were barely readable as the background
switches to dark for small widths.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
I do not found another way to disable inclusion in the sidebar...
The genindex information is alredy provided through glossary
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
restructured text comment syntax, i.e., everything it cannot parse is
a comment, is a real PITA!
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
avoid hardcoding width in the docs itself, so that other render
outputs can choose another size.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
It's not all perfect (yet) but way cleaner and simpler to use than
the sphinx one.
Custom do the scrolling for the fixed side bar and make some other
slight adjustments.
Main issue for now is that the "Developer Appendix" is always shown
in the navigation tree, but we only include that toctree for
devbuilds...
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Allows to differ the following situations:
* some snapshots in a group where not verified
* how many snapshots failed to verify in a group
* all snapshots verified but last verification task was over 30 days
ago
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Using 75 as width we can display up to 9999999 which would allow
displaying over 19 years of snapshots done each minute, so quite
enough for the common cases.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Because if not, the backups it creates have bogus permissions and may
seem like they got broken once the daemon is started again with the
correct user/group.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Double-click on the verify grid-cell of a specific snapshot (not the
group) opens the relevant task log.
The date of the last verify is shown as tool-tip.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Save the state ("ok" or "failed") and the UPID of the respective
verify task. With this we can easily allow to open the relevant task
log and show when the last verify happened.
As we already load the manifest when listing the snapshots, just add
it there directly.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
At the time when we can fix this up the new (and possibly an old)
server daemon process is running, so use the flock CLI tool from
util-linux to ensure we do the same locking as the server and thus we
avoid a race condition.
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
taken mostly from PVE, with adaption to how PBS does things.
Main difference is that we do not have a resource store singleton
here which we can use, but for datastores we can already use the
always present datastore-list store. Register it to the store manager
with a "storeId" property (vs. our internal storeid one).
Signed-off-by: Thomas Lamprecht <t.lamprecht@proxmox.com>
Thomas Lamprecht
