Date | Commit message (Collapse) |
|
This is no longer limited to Maildirs now that IMAP and NNTP
support exist; so give it a shorter name.
|
|
Quiet down logs from -imapd when clients are blindly
sending some unsupported flag conditions (e.g. "DRAFT",
"DELETED") specified in RFC 3501.
|
|
Link: https://public-inbox.org/meta/20200828221803.GA89978@dcvr/
|
|
We'll deduplicate redundant lines and show counts of skipped
tests to ensure it's easy to notice if something is unexpectedly
skipped.
|
|
There's no need for this to be a separate sub since there's
only a single caller. This saves a few kilobytes at least
in short-lived processes.
|
|
As noted in commit 87dca6d8d5988c5eb54019cca342450b0b7dd6b7
("www: rework query responses to avoid COUNT in SQLite"),
COUNT on many rows is expensive on big SQLite DBs.
We've already stopped using that code path long ago in WWW
while -imapd and -nntpd never used it. So we'll adjust our
remaining test cases to not need it, either.
|
|
Since we got rid of over->connect, `disconnect' no longer pairs
with it. So name it after the `close(2)' syscall it ultimately
issues.
|
|
`->connect' is confused with the perlfunc for the `connect(2)'
syscall, and also `DBI->connect'. Since SQLite doesn't use
sockets, the word "connect" needlessly confuses me. Give
it a short name to match the field name we use for it, which
also matches the variable name used by the DBI(3pm) and
DBD::SQLite(3pm) manpages.
|
|
WAL actually seems to have ideal locking characteristics given
concurrency problems I'm experiencing with --reindex running
in parallel with expensive read-only SQLite queries:
<https://public-inbox.org/meta/20200825001204.GA840@dcvr/>
Unfortunately, we cannot blindly use WAL while preserving
compatibility with existing setups nor our guarantees that
read-only daemons are indeed "read-only".
However, respect an user's the choice to set WAL on their
own if they're comfortable with giving -nntpd/-httpd/-imapd
processes write permission to the directory storing SQLite DBs.
|
|
Expanding threads via over.sqlite3 for mbox.gz downloads without
Xapian effectively collapsing on the THREADID column leads to
repeated messages getting downloaded.
To avoid that situation, use a "has_threadid" Xapian metadata
flag that's only set on --reindex (and brand new Xapian DBs).
This allows admins to upgrade WWW or do --reindex in any order;
without worrying about users eating up bandwidth and CPU cycles.
|
|
This is the `tid' column from over.sqlite3; and will be used for
IMAP and JMAP search (among other things).
|
|
Since we no longer read document data from Xapian, allow users
to opt-out of storing it.
This breaks compatibility with previous releases of
public-inbox, but gives us a ~1.5% space savings on Xapian
storage (and associated I/O and page cache pressure reduction).
|
|
Numbers are hard :<
|
|
We no longer read docdata.glass from anywhere in our code base.
Some adjustments were needed to t/search.t to deal with the
Xapian::WritableDatabase committing at different times, since
our ->query is avoided from PublicInbox::SearchIdx to avoid
needing a {over_ro} field.
|
|
It may be too easily confused for --newsgroup or --ng. This is
too rarely used and never made it into a release, so it should
be fine.
|
|
We can reduce the need to edit the config file for NNTP group names
this way.
|
|
There may be messages in the wild with wide characters in
headers which aren't non-RFC2047 encoded. Assume UTF-8 so
those fields can round trip through over.sqlite3.
This doesn't affect docdata.glass in Xapian, but it does
affect how over.sqlite3 stores the same deflated info.
|
|
We can use open(..., undef) natively in Perl in t/import.t
In places where we need a pathname, the File::Temp OO API
gives us auto-unlinking for free.
|
|
fileno(DIRHANDLE) only works on Perl 5.22+, so we need to use
dirfd(3) ourselves from Inline::C (or rely on chattr(1) being
installed).
While we're at it, rename `set_nodatacow' to `nodatacow_fd'
for consistency with `nodatacow_dir'.
|
|
This gives better page cache utilization for Xapian indexing on
slow storage by improving locality for random I/O activity on
the Xapian DB.
Instead of doing a single-pass to index both SQLite and Xapian;
this indexes them separately. The first pass is identical to
indexlevel=basic: it indexes both over.sqlite3 and msgmap.sqlite3.
Subsequent passes only operate on a single Xapian shard for
documents belonging to that shard. Given enough shards, each
individual shard can be made small enough to fit into the kernel
page cache and avoid HDD seeks for read activity.
Doing rough tests with a busy system with a 7200 RPM HDD with ext4,
full indexing of LKML (9 epochs) goes from ~80 hours (-j0) to
~30 hours (-j8) with 16GB RAM with 7 shards configured and fsync(2)
disabled (--no-sync) and `--batch-size=10m'.
|
|
I find myself mindlessly adding "-c" to public-inbox-index,
and other users may do the same. Instead of erroring out,
we'll just silently ignore it, for now and allow
public-inbox-compact to work on SQLite-only inboxes.
We'll only check for xapian-compact if search exists, since
it won't be needed in case we support SQLite VACUUM.
|
|
Thanks to the GCC compile farm project, we can wire up syscalls
for sparc64 and set system-specific SFD_* constants properly.
I've FINALLY figured out how to use POSIX::SigSet to generate
a usable buffer for the syscall perlfunc. This is required
for endian-neutral behavior and relevant to sparc64, at least.
There's no need for signalfd-related stuff to be constants,
either. signalfd initialization is never a hot path and a stub
subroutine for constants uses several KB of memory in the
interpreter.
We'll drop the needless SEEK_CUR import while we're importing
O_NONBLOCK, too.
|
|
This is specified in RFC 3501 but was accidentally omitted :x
I probably got it confused with TEXT, so add a comment about
TEXT being "everything" in the message.
|
|
While we always generate YYYYMMDDhhmmss query parameters
ourselves, the regexps in paginate_recent allow YYYYMMDD-only
(no hhmmss) timestamps, so don't trigger Time::Local::timegm
warnings about empty numeric comparisons on empty strings when a
client starts making up their own URLs.
|
|
epoll_wait_mod8 places a dummy element into the [2] slot of the
nested array, which caused is_deeply to fail.
Tested on aarch64.
|
|
We don't want ENV changes propagated to other tests
when using t/run.perl via "make check-run"
|
|
No need to waste resources when doing minimal work. With
PI_TEST_VERSION=2, this fixes a test failure where
Net::NNTP::DESTROY was getting called in the shard process.
We'll also get rid of an unnecessary use_ok under v2, too.
|
|
We can rely on the newer mids() sub directly and use faster
numeric comparisons for Msgmap unindexing in v1.
|
|
The return value of art_lookup changed but this command wasn't
updated since it wasn't tested.
Fixes: 0e6ceff37fc38f28 ("nntp: support slow blob retrievals")
|
|
v?fork failures seems to be the cause of locks not getting
released in -watch. Ensure lock release doesn't get skipped
in ->done for both v1 and v2 inboxes. We also need to do
everything we can to ensure DB handles, pipes and processes
get released even in the face of failure.
While we're at it, make failures around `git update-server-info'
non-fatal, since smart HTTP seems more popular anyways.
v2 changes:
- spawn: show failing command
- ensure waitpid is synchronous for inotify events
- teardown all fast-import processes on exception,
not just the failing one
- beef up lock_release error handling
- release lock on fast-import spawn failure
|
|
We'll just set the documented PI_EMERGENCY env to
a writable location.
|
|
The query parser test pulls in all of the IMAP stuff, so it
has the same dependencies.
|
|
SQLite and Xapian files are written randomly, thus they become
fragmented under btrfs with copy-on-write. This leads to
noticeable performance problems (and probably ENOSPC) as these
files get big.
lore/git (v2, <1GB) indexes around 20% faster with this on an
ancient SSD. lore/lkml seems to be taking forever and I'll
probably cancel it to save wear on my SSD.
Unfortunately, disabling CoW also means disabling checksumming
(and compression), so we'll be careful to only set the No_COW
attribute on regeneratable data. We want to keep CoW (and
checksums+compression) on git storage because current ref
storage is neither checksummed nor compressed, and git streams
pack output.
|
|
Otherwise, a user is more likely to remove the msgmap-XXXXXXXX
SQLite file from $TMPDIR and cause SQLite to error out.
|
|
Tests for failures should not leave junk temporary files lying
around in a users' ~/.public-inbox/.
On a side note, I'm not sure if PI_DIR is or was ever
necessary. It's never been documented, so perhaps
using $HOME for this is better...
|
|
We'll switch to using IdxStack here to ensure we get repeatable
results and ascending THREADIDs according to git chronology.
This means we'll need a two-pass reindex to index existing
messages before indexing new messages.
Since we no longer have a long-lived git-log process, we don't
have to worry about old Xapian referencing the git-log pipe
w/o FD_CLOEXEC, either.
|
|
The "xdb" prefix was inaccurate since it's used by
indexlevel=basic, which is Xapian-free. The '_' (underscore)
prefix was also wrong for a method which is called across
package boundaries.
|
|
This avoids pinning a potentially large chunk of memory from
`git-log --reverse' into RAM (or triggering less predictable
swap behavior). Instead it uses a contiguous temporary file
with a fixed-size record for every blob we'll need to index.
|
|
Older versions of public-inbox < 1.3.0 had subtly
different semantics around threading in some corner
cases. This switch (when combined with --reindex)
allows us to fix them by regenerating associations.
|
|
Noticed while reindexing a largish v2 inbox in parallel on an
SSD which required checkpointing and respawning shard workers.
Fixes: f06e84220e5566e7 ("over+msgmap: do not store filename after DBI->connect")
|
|
Since over.sqlite3 seems here to stay, we no longer need to do
Message-ID lookups against Xapian and can simply rely on the
docid <=> NNTP article number equivalancy SCHEMA_VERSION=15
gave us.
This rids us of the closure-using batch_do sub in the v1
code path and vastly simplifies both v1 and v2 unindexing.
|
|
The eval in key2sub via t/run.perl ("make check-run") won't
trigger the warning, but running "prove -bvw t/import.t"
directly, does. In any case, ensure the contents of this
variable doesn't linger across runs.
|
|
"\n" and other characters requiring quoting and/or escaping in
in $GIT_DIR/objects/info/alternates was not supported in git 2.11
and earlier; nor does it seem supported at all in libgit2.
This will allow us to support sharing git-cat-file or similar
endpoints across multiple inboxes via alternates.
This breaks an existing use case for anybody wacky
enough to put `\n' in the `inboxdir' pathname; but I doubt
this affects anybody.
|
|
SQLite already knows the filename internally, so avoid having it
as a long-lived Perl SV to save some bytes when there's many
inboxes and open DBs.
|
|
While it's even less common to experience a replaced
msgmap.sqlite3 file, BOFHs may do the darndest things. This is
another step towards reducing the number of needless wakeups
we need to do in long-lived read-only daemons.
|
|
Non-slice mailboxes never have messages themselves,
so we must not assume a message exists when sending
untagged EXISTS messages.
|
|
Returning an empty string for a filename makes no sense,
so instead return `undef' so the caller can setup a fallback
using the "//" operator.
This fixes uninitialized variable warnings because split()
on an empty string returns `undef', which caused to_filename
to warn on s// and tr// ops.
|
|
Since Perl doesn't internally use a self-pipe for
sleep/select/poll/etc, wake up every 10ms to ensure
it can see the SIGCHLD; since neither signalfd nor EVFILT_SIGNAL
are always available.
Fixes: 761baa2a300e4268 ("spawn: unblock SIGCHLD in subprocess")
|
|
We can reuse some of the GzipFilter infrastructure used by other
WWW components to handle slow blob retrieval, here. The
difference from previous changes is we don't decide on the 200
status code until we've retrieved the blob and found the
attachment.
While we're at it, ensure we can compress text attachment
responses once again, since all text attachments are served
as text/plain.
|
|
This allows -httpd to handle other requests while waiting
for git to retrieve and decode blobs. We'll also break
apart t/psgi_v2.t further to ensure tests run against
-httpd in addition to generic PSGI testing.
Using xt/httpd-async-stream.t to test against clones of meta@public-inbox.org
shows a 10-12% performance improvement with the following env:
TEST_JOBS=1000 TEST_CURL_OPT=--compressed TEST_ENDPOINT=new.atom
|